diff --git a/data/processed/RQ1/EmailResponder-MCP/evaluate_success.py b/data/processed/RQ1/EmailResponder-MCP/evaluate_success.py index 0abfdc506ade69a07d98e2900701496d216c3087..619cc5845948694a8276b146d52c102522cbdd9b 100644 --- a/data/processed/RQ1/EmailResponder-MCP/evaluate_success.py +++ b/data/processed/RQ1/EmailResponder-MCP/evaluate_success.py @@ -115,7 +115,7 @@ def save_detailed_results(results, output_file="success-finish_detailed_results. print(f"\n[OK] Detailed results saved to: {output_path}") -def save_csv_summary(results, output_file="success-finish_rate.csv"): +def save_csv_summary(results, output_file="success_rate.csv"): """Save summary to CSV file.""" import csv diff --git a/data/processed/RQ1/EmailResponder-MCP/success-finish_rate.csv b/data/processed/RQ1/EmailResponder-MCP/success_rate.csv similarity index 100% rename from data/processed/RQ1/EmailResponder-MCP/success-finish_rate.csv rename to data/processed/RQ1/EmailResponder-MCP/success_rate.csv diff --git a/data/processed/RQ1/EmailResponder/evaluate_success.py b/data/processed/RQ1/EmailResponder/evaluate_success.py index 0c19cad951953ca35b55035180011f9c663c2fe3..8f51414aeed16635d9a4cfa1807eb97e163e5752 100644 --- a/data/processed/RQ1/EmailResponder/evaluate_success.py +++ b/data/processed/RQ1/EmailResponder/evaluate_success.py @@ -114,7 +114,7 @@ def save_detailed_results(results, output_file="success-finish_detailed_results. print(f"\n✅ Detailed results saved to: {output_path}") -def save_csv_summary(results, output_file="success-finish_rate.csv"): +def save_csv_summary(results, output_file="success_rate.csv"): """Save summary to CSV file""" import csv diff --git a/data/processed/RQ1/EmailResponder/success-finish_rate.csv b/data/processed/RQ1/EmailResponder/success_rate.csv similarity index 100% rename from data/processed/RQ1/EmailResponder/success-finish_rate.csv rename to data/processed/RQ1/EmailResponder/success_rate.csv diff --git a/data/processed/RQ1/GameBuilder-MCP/evaluate_success.py b/data/processed/RQ1/GameBuilder-MCP/evaluate_success.py index ded983ecb1d20fb54ab3ec376d944d91d2ff6f49..927c3705fd9131f43a0b8eb34d7181e8eeca9b4e 100644 --- a/data/processed/RQ1/GameBuilder-MCP/evaluate_success.py +++ b/data/processed/RQ1/GameBuilder-MCP/evaluate_success.py @@ -115,7 +115,7 @@ def save_detailed_results(results, output_file="success-finish_detailed_results. print(f"\n✅ Detailed results saved to: {output_path}") -def save_csv_summary(results, output_file="success-finish_rate.csv"): +def save_csv_summary(results, output_file="success_rate.csv"): """Save a summary to a CSV file.""" import csv diff --git a/data/processed/RQ1/GameBuilder-MCP/success-finish_rate.csv b/data/processed/RQ1/GameBuilder-MCP/success_rate.csv similarity index 100% rename from data/processed/RQ1/GameBuilder-MCP/success-finish_rate.csv rename to data/processed/RQ1/GameBuilder-MCP/success_rate.csv diff --git a/data/processed/RQ1/GameBuilder/evaluate_success.py b/data/processed/RQ1/GameBuilder/evaluate_success.py index 8ca08338872947fcc8bc2b31b4dae548817ab03c..f0a55eb3e35114c6f677fd92121794572d8ae3c2 100644 --- a/data/processed/RQ1/GameBuilder/evaluate_success.py +++ b/data/processed/RQ1/GameBuilder/evaluate_success.py @@ -115,7 +115,7 @@ def save_detailed_results(results, output_file="success-finish_detailed_results. print(f"\n✅ Detailed results saved to: {output_path}") -def save_csv_summary(results, output_file="success-finish_rate.csv"): +def save_csv_summary(results, output_file="success_rate.csv"): """Save summary to CSV file""" import csv diff --git a/data/processed/RQ1/GameBuilder/success-finish_rate.csv b/data/processed/RQ1/GameBuilder/success_rate.csv similarity index 100% rename from data/processed/RQ1/GameBuilder/success-finish_rate.csv rename to data/processed/RQ1/GameBuilder/success_rate.csv diff --git a/data/processed/RQ1/MarkdownValidator-MCP/evaluate_success.py b/data/processed/RQ1/MarkdownValidator-MCP/evaluate_success.py index c1733e33728af53cf51aa838292ed7cdaaefb4f2..b626f32afebf741f7d925b283990ae7eab08173a 100644 --- a/data/processed/RQ1/MarkdownValidator-MCP/evaluate_success.py +++ b/data/processed/RQ1/MarkdownValidator-MCP/evaluate_success.py @@ -115,7 +115,7 @@ def save_detailed_results(results, output_file="success-finish_detailed_results. print(f"\nDetailed results saved to: {output_path}") -def save_csv_summary(results, output_file="success-finish_rate.csv"): +def save_csv_summary(results, output_file="success_rate.csv"): """Save summary to CSV file""" import csv diff --git a/data/processed/RQ1/MarkdownValidator-MCP/success-finish_rate.csv b/data/processed/RQ1/MarkdownValidator-MCP/success_rate.csv similarity index 100% rename from data/processed/RQ1/MarkdownValidator-MCP/success-finish_rate.csv rename to data/processed/RQ1/MarkdownValidator-MCP/success_rate.csv diff --git a/data/processed/RQ1/MarkdownValidator/evaluate_success.py b/data/processed/RQ1/MarkdownValidator/evaluate_success.py index 1887659e495e23dd91686866cfa92d750b374325..84c166ece75ab9ff762ba72ef24eadb8dce1969b 100644 --- a/data/processed/RQ1/MarkdownValidator/evaluate_success.py +++ b/data/processed/RQ1/MarkdownValidator/evaluate_success.py @@ -115,7 +115,7 @@ def save_detailed_results(results, output_file="success-finish_detailed_results. print(f"\nDetailed results saved: {output_path}") -def save_csv_summary(results, output_file="success-finish_rate.csv"): +def save_csv_summary(results, output_file="success_rate.csv"): """Save summary to CSV file""" import csv diff --git a/data/processed/RQ1/MarkdownValidator/success-finish_rate.csv b/data/processed/RQ1/MarkdownValidator/success_rate.csv similarity index 100% rename from data/processed/RQ1/MarkdownValidator/success-finish_rate.csv rename to data/processed/RQ1/MarkdownValidator/success_rate.csv diff --git a/data/processed/RQ1/RadarCharts/All_Open_vs_Closed_summaries.md b/data/processed/RQ1/RadarCharts/All_Open_vs_Closed_summaries.md index f5c6f978a08fa02309f41dc9f94c8187726f094a..eb0251e815b9d99769db84df775a69c299a480af 100644 --- a/data/processed/RQ1/RadarCharts/All_Open_vs_Closed_summaries.md +++ b/data/processed/RQ1/RadarCharts/All_Open_vs_Closed_summaries.md @@ -2,62 +2,62 @@ | Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | | --- | --- | --- | --- | --- | --- | --- | --- | -| open | 3 | 0.21 | 0.44 | 0.84 | 0.93 | 0.18 | 0.65 | -| closed | 4 | 0.13 | 0.24 | 0.91 | 0.81 | 0.12 | 0.56 | +| open | 3 | 0.27 | 0.47 | 0.86 | 0.93 | 0.16 | 0.99 | +| closed | 4 | 0.16 | 0.27 | 0.92 | 0.82 | 0.11 | 1.00 | # BookWriter Open vs Closed Models -| Series | Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | -| --- | --- | --- | --- | --- | --- | --- | --- | --- | -| BookWriter | open | 3 | 0.01 | 0.20 | 0.91 | 0.86 | 0.04 | 1.00 | -| BookWriter | closed | 4 | 0.00 | 0.00 | 0.94 | 0.66 | 0.00 | 1.00 | +| Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | +| --- | --- | --- | --- | --- | --- | --- | --- | +| open | 3 | 0.01 | 0.20 | 0.91 | 0.86 | 0.04 | 1.00 | +| closed | 4 | 0.00 | 0.00 | 0.94 | 0.66 | 0.00 | 1.00 | # EmailResponder Open vs Closed Models -| Series | Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | -| --- | --- | --- | --- | --- | --- | --- | --- | --- | -| EmailResponder | open | 3 | 0.07 | 0.11 | 0.92 | 0.87 | 0.06 | 0.00 | -| EmailResponder | closed | 4 | 0.12 | 0.14 | 0.98 | 0.82 | 0.05 | 0.00 | +| Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | +| --- | --- | --- | --- | --- | --- | --- | --- | +| open | 3 | 0.07 | 0.11 | 0.92 | 0.87 | 0.06 | 1.00 | +| closed | 4 | 0.12 | 0.14 | 0.98 | 0.82 | 0.05 | 1.00 | # GameBuilder Open vs Closed Models -| Series | Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | -| --- | --- | --- | --- | --- | --- | --- | --- | --- | -| GameBuilder | open | 3 | 0.89 | 0.97 | 0.98 | 0.99 | 0.00 | 0.00 | -| GameBuilder | closed | 4 | 0.47 | 0.52 | 0.99 | 0.89 | 0.01 | 0.00 | +| Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | +| --- | --- | --- | --- | --- | --- | --- | --- | +| open | 3 | 0.89 | 0.97 | 0.98 | 0.99 | 0.00 | 0.99 | +| closed | 4 | 0.47 | 0.52 | 0.99 | 0.89 | 0.01 | 1.00 | # LandingPageGenerator Open vs Closed Models -| Series | Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | average_score | -| --- | --- | --- | --- | --- | --- | --- | --- | --- | -| LandingPageGenerator | open | 3 | 0.11 | 0.53 | 0.83 | 0.97 | 0.04 | 0.54 | -| LandingPageGenerator | closed | 4 | 0.08 | 0.35 | 0.89 | 0.87 | 0.01 | 0.29 | +| Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | average_score | +| --- | --- | --- | --- | --- | --- | --- | --- | +| open | 3 | 0.11 | 0.53 | 0.83 | 0.97 | 0.04 | 0.54 | +| closed | 4 | 0.08 | 0.35 | 0.89 | 0.87 | 0.01 | 0.29 | # MarkdownValidator Open vs Closed Models -| Series | Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | average_score | -| --- | --- | --- | --- | --- | --- | --- | --- | --- | -| MarkdownValidator | open | 3 | 0.99 | 0.99 | 1.00 | 1.00 | 0.00 | 0.82 | -| MarkdownValidator | closed | 4 | 0.59 | 0.61 | 0.96 | 0.90 | 0.07 | 0.44 | +| Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | average_score | +| --- | --- | --- | --- | --- | --- | --- | --- | +| open | 3 | 0.99 | 0.99 | 1.00 | 1.00 | 0.00 | 0.82 | +| closed | 4 | 0.59 | 0.61 | 0.96 | 0.90 | 0.07 | 0.44 | # RecruitmentAssistant Open vs Closed Models -| Series | Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | -| --- | --- | --- | --- | --- | --- | --- | --- | --- | -| RecruitmentAssistant | open | 3 | 0.01 | 0.31 | 0.81 | 0.94 | 0.10 | 0.99 | -| RecruitmentAssistant | closed | 4 | 0.00 | 0.17 | 0.92 | 0.83 | 0.00 | 1.00 | +| Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | +| --- | --- | --- | --- | --- | --- | --- | --- | +| open | 3 | 0.01 | 0.31 | 0.81 | 0.94 | 0.10 | 0.99 | +| closed | 4 | 0.00 | 0.17 | 0.92 | 0.83 | 0.00 | 1.00 | # SQLAssistant Open vs Closed Models -| Series | Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | average_score | -| --- | --- | --- | --- | --- | --- | --- | --- | --- | -| SQLAssistant | open | 3 | 0.00 | 0.15 | 0.66 | 0.91 | 0.59 | 0.47 | -| SQLAssistant | closed | 4 | 0.00 | 0.06 | 0.79 | 0.83 | 0.46 | 0.38 | +| Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | average_score | +| --- | --- | --- | --- | --- | --- | --- | --- | +| open | 3 | 0.00 | 0.15 | 0.66 | 0.91 | 0.59 | 0.47 | +| closed | 4 | 0.00 | 0.06 | 0.79 | 0.83 | 0.46 | 0.38 | # SocialMediaManager Open vs Closed Models -| Series | Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | -| --- | --- | --- | --- | --- | --- | --- | --- | --- | -| SocialMediaManager | open | 3 | 0.06 | 0.50 | 0.75 | 0.92 | 0.45 | 0.98 | -| SocialMediaManager | closed | 4 | 0.03 | 0.27 | 0.87 | 0.72 | 0.29 | 0.99 | +| Group | Model Count | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | +| --- | --- | --- | --- | --- | --- | --- | --- | +| open | 3 | 0.06 | 0.50 | 0.75 | 0.92 | 0.45 | 0.98 | +| closed | 4 | 0.03 | 0.27 | 0.87 | 0.72 | 0.29 | 0.99 | diff --git a/data/processed/RQ1/RadarCharts/All_Series_Average_tables.md b/data/processed/RQ1/RadarCharts/All_Series_Average_tables.md index 33386e42cd6df654c59beb8e0d828cb8415f07ff..57d7b8c5bfef8c7143fadcb8f6df2be219916478 100644 --- a/data/processed/RQ1/RadarCharts/All_Series_Average_tables.md +++ b/data/processed/RQ1/RadarCharts/All_Series_Average_tables.md @@ -1,107 +1,125 @@ -# Overall Summary (All Projects) +# Overall Summary (All Projects - Weighted Average) -| model | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | -| --- | --- | --- | --- | --- | --- | --- | -| GPT-5 | 0.24 | 0.31 | 0.93 | 0.90 | 0.11 | 0.63 | -| GPT-4o-mini | 0.24 | 0.61 | 0.83 | 0.95 | 0.12 | 0.67 | -| DeepSeek-V3-1 | 0.26 | 0.64 | 0.86 | 0.96 | 0.13 | 0.68 | -| DeepSeek-R1 | 0.18 | 0.26 | 0.79 | 0.90 | 0.25 | 0.59 | -| Gemini-2.5-flash | 0.03 | 0.03 | 0.93 | 0.69 | 0.14 | 0.48 | -| Gemini-2.5-flash-nothinking | 0.02 | 0.02 | 0.94 | 0.70 | 0.12 | 0.48 | -| Qwen3-235b | 0.20 | 0.41 | 0.88 | 0.92 | 0.16 | 0.67 | +_Note: average_score is only computed for GT projects (MarkdownValidator, SQLAssistant, LandingPageGenerator). pass_rate is only computed for non-GT projects._ + +| model | exact_match | any_order_match | precision | recall | retry_rate | average_score | pass_rate | +| --- | --- | --- | --- | --- | --- | --- | --- | +| DeepSeek-R1 | 0.13 | 0.21 | 0.77 | 0.90 | 0.26 | 0.45 | 0.98 | +| DeepSeek-V3-1 | 0.22 | 0.59 | 0.86 | 0.96 | 0.13 | 0.67 | 1.00 | +| GPT-4o-mini | 0.22 | 0.58 | 0.84 | 0.95 | 0.12 | 0.65 | 0.99 | +| GPT-5 | 0.18 | 0.25 | 0.93 | 0.89 | 0.11 | 0.52 | 1.00 | +| Gemini-2.5-flash | 0.03 | 0.03 | 0.93 | 0.69 | 0.14 | 0.13 | 1.00 | +| Gemini-2.5-flash-nothinking | 0.02 | 0.02 | 0.94 | 0.70 | 0.12 | 0.14 | 1.00 | +| Qwen3-235b | 0.16 | 0.36 | 0.88 | 0.91 | 0.16 | 0.64 | 0.99 | # BookWriter Series Average +_This is a non-GT project series. Using pass_rate from success_rate.csv._ + | model | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | | --- | --- | --- | --- | --- | --- | --- | -| GPT-5 | 0.00 | 0.00 | 0.97 | 0.77 | 0.00 | 1.00 | -| GPT-4o-mini | 0.00 | 0.00 | 0.85 | 0.83 | 0.00 | 1.00 | -| DeepSeek-V3-1 | 0.04 | 0.58 | 0.89 | 0.98 | 0.01 | 1.00 | | DeepSeek-R1 | 0.00 | 0.01 | 0.88 | 0.81 | 0.09 | 1.00 | +| DeepSeek-V3-1 | 0.04 | 0.58 | 0.89 | 0.98 | 0.01 | 1.00 | +| GPT-4o-mini | 0.00 | 0.00 | 0.85 | 0.83 | 0.00 | 1.00 | +| GPT-5 | 0.00 | 0.00 | 0.97 | 0.77 | 0.00 | 1.00 | | Gemini-2.5-flash | 0.00 | 0.00 | 0.98 | 0.52 | 0.00 | 1.00 | | Gemini-2.5-flash-nothinking | 0.00 | 0.00 | 0.98 | 0.53 | 0.01 | 1.00 | | Qwen3-235b | 0.00 | 0.00 | 0.95 | 0.79 | 0.01 | 1.00 | # EmailResponder Series Average +_This is a non-GT project series. Using pass_rate from success_rate.csv._ + | model | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | | --- | --- | --- | --- | --- | --- | --- | -| GPT-5 | 0.01 | 0.01 | 1.00 | 0.83 | 0.00 | 0.00 | -| GPT-4o-mini | 0.43 | 0.47 | 0.99 | 0.91 | 0.00 | 0.00 | -| DeepSeek-V3-1 | 0.16 | 0.24 | 0.98 | 0.88 | 0.02 | 0.00 | -| DeepSeek-R1 | 0.00 | 0.03 | 0.77 | 0.89 | 0.16 | 0.00 | -| Gemini-2.5-flash | 0.01 | 0.01 | 0.97 | 0.74 | 0.10 | 0.00 | -| Gemini-2.5-flash-nothinking | 0.05 | 0.06 | 0.96 | 0.81 | 0.12 | 0.00 | -| Qwen3-235b | 0.06 | 0.07 | 1.00 | 0.85 | 0.00 | 0.00 | +| DeepSeek-R1 | 0.00 | 0.03 | 0.77 | 0.89 | 0.16 | 0.99 | +| DeepSeek-V3-1 | 0.16 | 0.24 | 0.98 | 0.88 | 0.02 | 1.00 | +| GPT-4o-mini | 0.43 | 0.47 | 0.99 | 0.91 | 0.00 | 1.00 | +| GPT-5 | 0.01 | 0.01 | 1.00 | 0.83 | 0.00 | 1.00 | +| Gemini-2.5-flash | 0.01 | 0.01 | 0.97 | 0.74 | 0.10 | 0.99 | +| Gemini-2.5-flash-nothinking | 0.04 | 0.06 | 0.96 | 0.81 | 0.12 | 1.00 | +| Qwen3-235b | 0.06 | 0.07 | 1.00 | 0.85 | 0.00 | 1.00 | # GameBuilder Series Average +_This is a non-GT project series. Using pass_rate from success_rate.csv._ + | model | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | | --- | --- | --- | --- | --- | --- | --- | -| GPT-5 | 1.00 | 1.00 | 1.00 | 1.00 | 0.00 | 0.00 | -| GPT-4o-mini | 0.88 | 1.00 | 0.96 | 1.00 | 0.00 | 0.00 | -| DeepSeek-V3-1 | 0.90 | 1.00 | 0.97 | 1.00 | 0.00 | 0.00 | -| DeepSeek-R1 | 0.94 | 0.96 | 1.00 | 0.99 | 0.00 | 0.00 | -| Gemini-2.5-flash | 0.00 | 0.00 | 1.00 | 0.75 | 0.00 | 0.00 | -| Gemini-2.5-flash-nothinking | 0.02 | 0.06 | 0.99 | 0.80 | 0.04 | 0.00 | -| Qwen3-235b | 0.84 | 0.94 | 0.99 | 0.99 | 0.00 | 0.00 | +| DeepSeek-R1 | 0.94 | 0.96 | 1.00 | 0.99 | 0.00 | 1.00 | +| DeepSeek-V3-1 | 0.90 | 1.00 | 0.97 | 1.00 | 0.00 | 0.98 | +| GPT-4o-mini | 0.88 | 1.00 | 0.96 | 1.00 | 0.00 | 1.00 | +| GPT-5 | 1.00 | 1.00 | 1.00 | 1.00 | 0.00 | 1.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | 1.00 | 0.75 | 0.00 | 1.00 | +| Gemini-2.5-flash-nothinking | 0.02 | 0.06 | 0.99 | 0.80 | 0.04 | 1.00 | +| Qwen3-235b | 0.84 | 0.94 | 0.99 | 0.99 | 0.00 | 1.00 | # LandingPageGenerator Series Average +_This is a GT project series. Using average_score from score_summary.csv._ + | model | exact_match | any_order_match | precision | recall | retry_rate | average_score | | --- | --- | --- | --- | --- | --- | --- | -| GPT-5 | 0.29 | 0.42 | 0.86 | 0.97 | 0.00 | 0.47 | -| GPT-4o-mini | 0.04 | 0.98 | 0.78 | 1.00 | 0.00 | 0.67 | -| DeepSeek-V3-1 | 0.24 | 0.99 | 0.86 | 1.00 | 0.01 | 0.86 | | DeepSeek-R1 | 0.00 | 0.13 | 0.78 | 0.94 | 0.06 | 0.25 | +| DeepSeek-V3-1 | 0.24 | 0.99 | 0.86 | 1.00 | 0.01 | 0.86 | +| GPT-4o-mini | 0.04 | 0.98 | 0.78 | 1.00 | 0.00 | 0.67 | +| GPT-5 | 0.29 | 0.42 | 0.86 | 0.97 | 0.00 | 0.47 | | Gemini-2.5-flash | 0.00 | 0.00 | 0.97 | 0.76 | 0.00 | 0.00 | | Gemini-2.5-flash-nothinking | 0.00 | 0.00 | 0.95 | 0.77 | 0.03 | 0.01 | | Qwen3-235b | 0.09 | 0.45 | 0.84 | 0.97 | 0.03 | 0.52 | # MarkdownValidator Series Average +_This is a GT project series. Using average_score from score_summary.csv._ + | model | exact_match | any_order_match | precision | recall | retry_rate | average_score | | --- | --- | --- | --- | --- | --- | --- | -| GPT-5 | 1.00 | 1.00 | 1.00 | 1.00 | 0.00 | 0.71 | -| GPT-4o-mini | 0.97 | 1.00 | 0.99 | 1.00 | 0.00 | 0.82 | -| DeepSeek-V3-1 | 1.00 | 1.00 | 1.00 | 1.00 | 0.00 | 0.84 | | DeepSeek-R1 | 0.98 | 1.00 | 1.00 | 1.00 | 0.00 | 0.81 | +| DeepSeek-V3-1 | 1.00 | 1.00 | 1.00 | 1.00 | 0.00 | 0.84 | +| GPT-4o-mini | 0.97 | 1.00 | 0.99 | 1.00 | 0.00 | 0.82 | +| GPT-5 | 1.00 | 1.00 | 1.00 | 1.00 | 0.00 | 0.71 | | Gemini-2.5-flash | 0.26 | 0.33 | 0.87 | 0.88 | 0.21 | 0.08 | | Gemini-2.5-flash-nothinking | 0.13 | 0.13 | 0.96 | 0.73 | 0.06 | 0.14 | | Qwen3-235b | 0.98 | 0.98 | 1.00 | 0.99 | 0.00 | 0.81 | # RecruitmentAssistant Series Average +_This is a non-GT project series. Using pass_rate from success_rate.csv._ + | model | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | | --- | --- | --- | --- | --- | --- | --- | -| GPT-5 | 0.00 | 0.00 | 0.98 | 0.90 | 0.00 | 1.00 | -| GPT-4o-mini | 0.00 | 0.70 | 0.76 | 0.98 | 0.00 | 1.00 | -| DeepSeek-V3-1 | 0.00 | 0.36 | 0.84 | 0.97 | 0.00 | 1.00 | | DeepSeek-R1 | 0.00 | 0.11 | 0.84 | 0.90 | 0.05 | 0.99 | +| DeepSeek-V3-1 | 0.00 | 0.36 | 0.84 | 0.97 | 0.00 | 1.00 | +| GPT-4o-mini | 0.00 | 0.70 | 0.76 | 0.98 | 0.00 | 1.00 | +| GPT-5 | 0.00 | 0.00 | 0.98 | 0.90 | 0.00 | 1.00 | | Gemini-2.5-flash | 0.00 | 0.00 | 0.98 | 0.69 | 0.00 | 1.00 | | Gemini-2.5-flash-nothinking | 0.00 | 0.00 | 0.97 | 0.74 | 0.00 | 1.00 | | Qwen3-235b | 0.02 | 0.46 | 0.75 | 0.96 | 0.26 | 0.98 | # SQLAssistant Series Average +_This is a GT project series. Using average_score from score_summary.csv._ + | model | exact_match | any_order_match | precision | recall | retry_rate | average_score | | --- | --- | --- | --- | --- | --- | --- | -| GPT-5 | 0.00 | 0.01 | 0.82 | 0.86 | 0.44 | 0.45 | -| GPT-4o-mini | 0.00 | 0.22 | 0.77 | 0.94 | 0.37 | 0.52 | -| DeepSeek-V3-1 | 0.00 | 0.17 | 0.66 | 0.90 | 0.61 | 0.37 | | DeepSeek-R1 | 0.00 | 0.04 | 0.52 | 0.88 | 0.80 | 0.40 | +| DeepSeek-V3-1 | 0.00 | 0.17 | 0.66 | 0.90 | 0.61 | 0.37 | +| GPT-4o-mini | 0.00 | 0.22 | 0.77 | 0.94 | 0.37 | 0.52 | +| GPT-5 | 0.00 | 0.01 | 0.82 | 0.86 | 0.44 | 0.45 | | Gemini-2.5-flash | 0.00 | 0.00 | 0.76 | 0.77 | 0.57 | 0.29 | | Gemini-2.5-flash-nothinking | 0.00 | 0.00 | 0.82 | 0.76 | 0.48 | 0.27 | | Qwen3-235b | 0.00 | 0.24 | 0.80 | 0.95 | 0.37 | 0.64 | # SocialMediaManager Series Average +_This is a non-GT project series. Using pass_rate from success_rate.csv._ + | model | exact_match | any_order_match | precision | recall | retry_rate | pass_rate | | --- | --- | --- | --- | --- | --- | --- | -| GPT-5 | 0.02 | 0.37 | 0.86 | 0.88 | 0.34 | 0.98 | -| GPT-4o-mini | 0.12 | 0.71 | 0.72 | 0.97 | 0.47 | 0.96 | -| DeepSeek-V3-1 | 0.17 | 0.87 | 0.78 | 0.97 | 0.27 | 1.00 | | DeepSeek-R1 | 0.01 | 0.22 | 0.64 | 0.88 | 0.62 | 0.94 | +| DeepSeek-V3-1 | 0.17 | 0.87 | 0.78 | 0.97 | 0.27 | 1.00 | +| GPT-4o-mini | 0.12 | 0.71 | 0.72 | 0.97 | 0.47 | 0.96 | +| GPT-5 | 0.02 | 0.37 | 0.86 | 0.88 | 0.34 | 0.98 | | Gemini-2.5-flash | 0.00 | 0.00 | 0.94 | 0.52 | 0.21 | 1.00 | | Gemini-2.5-flash-nothinking | 0.00 | 0.00 | 0.96 | 0.51 | 0.15 | 1.00 | | Qwen3-235b | 0.02 | 0.42 | 0.82 | 0.89 | 0.46 | 0.98 | diff --git a/data/processed/RQ1/RadarCharts/BookWriter-A2A_radar.pdf b/data/processed/RQ1/RadarCharts/BookWriter-A2A_radar.pdf index 50d48944767e2b64855e11859418be548b3f943e..f88e1de0ce141c2549eeb1916be7a3342846a35c 100644 --- a/data/processed/RQ1/RadarCharts/BookWriter-A2A_radar.pdf +++ b/data/processed/RQ1/RadarCharts/BookWriter-A2A_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:9eaed441462082800ee22bfb8ef5a1cc84ab4c7f9a1a956570ad6fe986e7034d -size 18128 +oid sha256:4da0e9d550b5ba06e37a72a279ac4fbdd6a2a4cda6ff704d37aa17907569d41c +size 18417 diff --git a/data/processed/RQ1/RadarCharts/BookWriter-H_A2A_radar.pdf b/data/processed/RQ1/RadarCharts/BookWriter-H_A2A_radar.pdf index 9d186e612d4f447b7821c8a234d44cb849978e0c..5546b61ef47edb47f0fabeea64c59cfd58083ac6 100644 --- a/data/processed/RQ1/RadarCharts/BookWriter-H_A2A_radar.pdf +++ b/data/processed/RQ1/RadarCharts/BookWriter-H_A2A_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:747927181126917ca0e6e331f60b03c32440bcaf92d46c0f5bc6e884b8b81d37 -size 18037 +oid sha256:27a597245cd6396c7f09be84845e95571d6207e13c19c51dfd12287a02d6eab8 +size 18321 diff --git a/data/processed/RQ1/RadarCharts/BookWriter-MCP_radar.pdf b/data/processed/RQ1/RadarCharts/BookWriter-MCP_radar.pdf index 1be4ecf9e77a7cec74e28eb2c6bc798214f4ab45..591ec51fe79157d0ae63329e71e3fefcffa6af99 100644 --- a/data/processed/RQ1/RadarCharts/BookWriter-MCP_radar.pdf +++ b/data/processed/RQ1/RadarCharts/BookWriter-MCP_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:1df13bbe21a802b24c81eaeb1884f69994df008f757f94abf545611e0cb94001 -size 18171 +oid sha256:c212b19b2b232560f4be82708dcc3627d4fd1672f1a6f17bf706fc4b84685fae +size 18476 diff --git a/data/processed/RQ1/RadarCharts/BookWriter_Series_Average_radar.pdf b/data/processed/RQ1/RadarCharts/BookWriter_Series_Average_radar.pdf index 30af835e0f3ca78fbcf18216f957098d1e8e34df..7d6351ae945a0e818cf9b442c712fa202628ca95 100644 --- a/data/processed/RQ1/RadarCharts/BookWriter_Series_Average_radar.pdf +++ b/data/processed/RQ1/RadarCharts/BookWriter_Series_Average_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:8b72a4c02471ff93ea36ece48d9907dafba89e32d9d5e15387c1d4ca000d45e9 +oid sha256:79c4c09f9bf7c088d0604cc23cad260c52eb608b0033805f0664ab62d0e8d3dc size 18189 diff --git a/data/processed/RQ1/RadarCharts/Comparison_A2A_vs_A2A_mix.md b/data/processed/RQ1/RadarCharts/Comparison_A2A_vs_A2A_mix.md deleted file mode 100644 index 2b596862a93dc0ce6ebf285cf13d58262335fb10..0000000000000000000000000000000000000000 --- a/data/processed/RQ1/RadarCharts/Comparison_A2A_vs_A2A_mix.md +++ /dev/null @@ -1,68 +0,0 @@ -# Overall A2A vs A2A_mix Comparison - -Averaged across all projects: SQL_assistant, intelligent_recruitment_platform, landing_page_generator, self_evaluation_loop_flow, write_a_book_with_flows - -## Framework-Level Comparison (All Models Averaged) - -| Metric | A2A | A2A_mix | Diff (A2A-A2A_mix) | -| --- | --- | --- | --- | - ---- - -## exact_match - -| Model | A2A (Avg) | A2A_mix (Avg) | Diff (A2A-A2A_mix) | -| --- | --- | --- | --- | - -## any_order_match - -| Model | A2A (Avg) | A2A_mix (Avg) | Diff (A2A-A2A_mix) | -| --- | --- | --- | --- | - -## precision - -| Model | A2A (Avg) | A2A_mix (Avg) | Diff (A2A-A2A_mix) | -| --- | --- | --- | --- | - -## recall - -| Model | A2A (Avg) | A2A_mix (Avg) | Diff (A2A-A2A_mix) | -| --- | --- | --- | --- | - -## retry_rate - -| Model | A2A (Avg) | A2A_mix (Avg) | Diff (A2A-A2A_mix) | -| --- | --- | --- | --- | - -## average_score - -| Model | A2A (Avg) | A2A_mix (Avg) | Diff (A2A-A2A_mix) | -| --- | --- | --- | --- | - -## pass_rate - -| Model | A2A (Avg) | A2A_mix (Avg) | Diff (A2A-A2A_mix) | -| --- | --- | --- | --- | - ---- - -# SQL_assistant: A2A vs A2A_mix - -_Data not available for comparison_ - -# intelligent_recruitment_platform: A2A vs A2A_mix - -_Data not available for comparison_ - -# landing_page_generator: A2A vs A2A_mix - -_Data not available for comparison_ - -# self_evaluation_loop_flow: A2A vs A2A_mix - -_Data not available for comparison_ - -# write_a_book_with_flows: A2A vs A2A_mix - -_Data not available for comparison_ - diff --git a/data/processed/RQ1/RadarCharts/Comparison_A2A_vs_H_A2A.md b/data/processed/RQ1/RadarCharts/Comparison_A2A_vs_H_A2A.md new file mode 100644 index 0000000000000000000000000000000000000000..c1e7d56ddded799d6fef2406000dfa8bd634da90 --- /dev/null +++ b/data/processed/RQ1/RadarCharts/Comparison_A2A_vs_H_A2A.md @@ -0,0 +1,539 @@ +# Overall A2A vs H_A2A Comparison + +Averaged across all projects: BookWriter, LandingPageGenerator, RecruitmentAssistant, SQLAssistant, SocialMediaManager + +## Framework-Level Comparison (All Models Averaged) + +| Metric | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| exact_match | 0.01 | 0.01 | -0.01 | +| any_order_match | 0.25 | 0.19 | +0.05 | +| precision | 0.82 | 0.85 | -0.03 | +| recall | 0.84 | 0.87 | -0.03 | +| retry_rate | 0.22 | 0.14 | +0.08 | +| average_score | 0.31 | 0.45 | -0.14 | +| pass_rate | 0.99 | 1.00 | -0.01 | + +--- + +## exact_match + +| Model | A2A (Avg) | H_A2A (Avg) | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.00 | 0.00 | -0.00 | +| DeepSeek-V3-1 | 0.03 | 0.04 | -0.01 | +| GPT-4o-mini | 0.01 | 0.05 | -0.05 | +| GPT-5 | 0.01 | 0.00 | +0.01 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.01 | 0.01 | +0.00 | + +## any_order_match + +| Model | A2A (Avg) | H_A2A (Avg) | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.05 | 0.12 | -0.07 | +| DeepSeek-V3-1 | 0.48 | 0.68 | -0.20 | +| GPT-4o-mini | 0.60 | 0.43 | +0.16 | +| GPT-5 | 0.17 | 0.03 | +0.14 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.43 | 0.10 | +0.33 | + +## precision + +| Model | A2A (Avg) | H_A2A (Avg) | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.70 | 0.79 | -0.09 | +| DeepSeek-V3-1 | 0.74 | 0.84 | -0.11 | +| GPT-4o-mini | 0.79 | 0.81 | -0.03 | +| GPT-5 | 0.89 | 0.91 | -0.02 | +| Gemini-2.5-flash | 0.91 | 0.89 | +0.02 | +| Gemini-2.5-flash-nothinking | 0.93 | 0.90 | +0.03 | +| Qwen3-235b | 0.81 | 0.83 | -0.02 | + +## recall + +| Model | A2A (Avg) | H_A2A (Avg) | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.88 | 0.89 | -0.00 | +| DeepSeek-V3-1 | 0.95 | 0.96 | -0.01 | +| GPT-4o-mini | 0.96 | 0.92 | +0.04 | +| GPT-5 | 0.90 | 0.85 | +0.04 | +| Gemini-2.5-flash | 0.62 | 0.78 | -0.16 | +| Gemini-2.5-flash-nothinking | 0.64 | 0.78 | -0.14 | +| Qwen3-235b | 0.93 | 0.91 | +0.02 | + +## retry_rate + +| Model | A2A (Avg) | H_A2A (Avg) | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.37 | 0.26 | +0.11 | +| DeepSeek-V3-1 | 0.25 | 0.10 | +0.15 | +| GPT-4o-mini | 0.15 | 0.09 | +0.07 | +| GPT-5 | 0.18 | 0.06 | +0.11 | +| Gemini-2.5-flash | 0.19 | 0.18 | +0.01 | +| Gemini-2.5-flash-nothinking | 0.18 | 0.12 | +0.06 | +| Qwen3-235b | 0.25 | 0.20 | +0.05 | + +## average_score + +| Model | A2A (Avg) | H_A2A (Avg) | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.16 | 0.31 | -0.15 | +| DeepSeek-V3-1 | 0.49 | 0.70 | -0.21 | +| GPT-4o-mini | 0.57 | 0.58 | -0.01 | +| GPT-5 | 0.33 | 0.42 | -0.09 | +| Gemini-2.5-flash | 0.08 | 0.27 | -0.20 | +| Gemini-2.5-flash-nothinking | 0.03 | 0.29 | -0.26 | +| Qwen3-235b | 0.49 | 0.55 | -0.06 | + +## pass_rate + +| Model | A2A (Avg) | H_A2A (Avg) | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.97 | 0.99 | -0.02 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 0.98 | 1.00 | -0.02 | +| GPT-5 | 0.99 | 1.00 | -0.01 | +| Gemini-2.5-flash | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 1.00 | 1.00 | +0.00 | +| Qwen3-235b | 0.99 | 0.99 | -0.01 | + +--- + +# BookWriter: A2A vs H_A2A + +## Overall Summary (Averaged Across All Models) + +| Metric | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| exact_match | 0.00 | 0.00 | +0.00 | +| any_order_match | 0.08 | 0.08 | -0.00 | +| precision | 0.93 | 0.91 | +0.02 | +| recall | 0.77 | 0.73 | +0.03 | +| retry_rate | 0.02 | 0.00 | +0.02 | +| pass_rate | 1.00 | 1.00 | +0.00 | + +--- + +## exact_match + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | +| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.00 | 0.00 | +0.00 | + +## any_order_match + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.02 | 0.00 | +0.02 | +| DeepSeek-V3-1 | 0.53 | 0.58 | -0.05 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.00 | 0.00 | +0.00 | + +## precision + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.87 | 0.91 | -0.03 | +| DeepSeek-V3-1 | 0.79 | 0.92 | -0.13 | +| GPT-4o-mini | 0.90 | 0.80 | +0.10 | +| GPT-5 | 0.98 | 0.95 | +0.03 | +| Gemini-2.5-flash | 1.00 | 0.93 | +0.07 | +| Gemini-2.5-flash-nothinking | 1.00 | 0.93 | +0.07 | +| Qwen3-235b | 0.96 | 0.93 | +0.03 | + +## recall + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.84 | 0.77 | +0.07 | +| DeepSeek-V3-1 | 0.99 | 0.95 | +0.04 | +| GPT-4o-mini | 0.86 | 0.77 | +0.09 | +| GPT-5 | 0.82 | 0.67 | +0.15 | +| Gemini-2.5-flash | 0.53 | 0.57 | -0.05 | +| Gemini-2.5-flash-nothinking | 0.53 | 0.57 | -0.04 | +| Qwen3-235b | 0.80 | 0.83 | -0.04 | + +## retry_rate + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.13 | 0.02 | +0.12 | +| DeepSeek-V3-1 | 0.02 | 0.00 | +0.02 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.00 | 0.00 | +0.00 | + +## pass_rate + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 1.00 | 1.00 | +0.00 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 1.00 | 1.00 | +0.00 | +| GPT-5 | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 1.00 | 1.00 | +0.00 | +| Qwen3-235b | 1.00 | 1.00 | +0.00 | + +# LandingPageGenerator: A2A vs H_A2A + +## Overall Summary (Averaged Across All Models) + +| Metric | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| exact_match | 0.00 | 0.00 | +0.00 | +| any_order_match | 0.40 | 0.29 | +0.11 | +| precision | 0.84 | 0.82 | +0.03 | +| recall | 0.91 | 0.93 | -0.02 | +| retry_rate | 0.02 | 0.01 | +0.01 | +| average_score | 0.30 | 0.33 | -0.03 | + +--- + +## exact_match + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | +| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.00 | 0.00 | +0.00 | + +## any_order_match + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.05 | 0.03 | +0.02 | +| DeepSeek-V3-1 | 1.00 | 0.98 | +0.02 | +| GPT-4o-mini | 0.97 | 1.00 | -0.03 | +| GPT-5 | 0.27 | 0.00 | +0.27 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.48 | 0.00 | +0.48 | + +## precision + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.76 | 0.78 | -0.03 | +| DeepSeek-V3-1 | 0.83 | 0.77 | +0.06 | +| GPT-4o-mini | 0.74 | 0.78 | -0.03 | +| GPT-5 | 0.81 | 0.79 | +0.02 | +| Gemini-2.5-flash | 1.00 | 0.91 | +0.09 | +| Gemini-2.5-flash-nothinking | 0.98 | 0.87 | +0.12 | +| Qwen3-235b | 0.79 | 0.81 | -0.02 | + +## recall + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.94 | 0.93 | +0.01 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 1.00 | 1.00 | -0.00 | +| GPT-5 | 0.97 | 0.94 | +0.03 | +| Gemini-2.5-flash | 0.74 | 0.87 | -0.13 | +| Gemini-2.5-flash-nothinking | 0.76 | 0.85 | -0.08 | +| Qwen3-235b | 0.98 | 0.94 | +0.04 | + +## retry_rate + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.03 | 0.00 | +0.03 | +| DeepSeek-V3-1 | 0.00 | 0.03 | -0.03 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.05 | 0.00 | +0.05 | +| Qwen3-235b | 0.05 | 0.05 | +0.00 | + +## average_score + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.08 | 0.10 | -0.02 | +| DeepSeek-V3-1 | 0.86 | 0.88 | -0.02 | +| GPT-4o-mini | 0.57 | 0.59 | -0.02 | +| GPT-5 | 0.19 | 0.32 | -0.12 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.38 | 0.41 | -0.03 | + +# RecruitmentAssistant: A2A vs H_A2A + +## Overall Summary (Averaged Across All Models) + +| Metric | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| exact_match | 0.00 | 0.00 | +0.00 | +| any_order_match | 0.18 | 0.31 | -0.13 | +| precision | 0.89 | 0.87 | +0.02 | +| recall | 0.86 | 0.95 | -0.09 | +| retry_rate | 0.05 | 0.03 | +0.01 | +| pass_rate | 1.00 | 1.00 | -0.00 | + +--- + +## exact_match + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | +| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.02 | 0.00 | +0.02 | + +## any_order_match + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.00 | 0.25 | -0.25 | +| DeepSeek-V3-1 | 0.00 | 1.00 | -1.00 | +| GPT-4o-mini | 0.68 | 0.72 | -0.03 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.57 | 0.22 | +0.35 | + +## precision + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.83 | 0.89 | -0.06 | +| DeepSeek-V3-1 | 0.87 | 0.82 | +0.05 | +| GPT-4o-mini | 0.79 | 0.75 | +0.05 | +| GPT-5 | 0.99 | 0.97 | +0.02 | +| Gemini-2.5-flash | 1.00 | 0.94 | +0.06 | +| Gemini-2.5-flash-nothinking | 1.00 | 0.93 | +0.06 | +| Qwen3-235b | 0.74 | 0.76 | -0.02 | + +## recall + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.88 | 0.93 | -0.06 | +| DeepSeek-V3-1 | 0.96 | 1.00 | -0.04 | +| GPT-4o-mini | 0.98 | 0.99 | -0.00 | +| GPT-5 | 0.90 | 0.91 | -0.01 | +| Gemini-2.5-flash | 0.64 | 0.89 | -0.25 | +| Gemini-2.5-flash-nothinking | 0.68 | 0.94 | -0.26 | +| Qwen3-235b | 0.97 | 0.96 | +0.01 | + +## retry_rate + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.07 | 0.05 | +0.02 | +| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.27 | 0.18 | +0.08 | + +## pass_rate + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.98 | 1.00 | -0.02 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 1.00 | 1.00 | +0.00 | +| GPT-5 | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 1.00 | 1.00 | +0.00 | +| Qwen3-235b | 0.98 | 1.00 | -0.02 | + +# SQLAssistant: A2A vs H_A2A + +## Overall Summary (Averaged Across All Models) + +| Metric | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| exact_match | 0.00 | 0.00 | +0.00 | +| any_order_match | 0.15 | 0.00 | +0.15 | +| precision | 0.65 | 0.83 | -0.18 | +| recall | 0.87 | 0.87 | -0.01 | +| retry_rate | 0.65 | 0.38 | +0.27 | +| average_score | 0.32 | 0.57 | -0.25 | + +--- + +## exact_match + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | +| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.00 | 0.00 | +0.00 | + +## any_order_match + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.03 | 0.00 | +0.03 | +| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | +| GPT-4o-mini | 0.53 | 0.00 | +0.53 | +| GPT-5 | 0.03 | 0.00 | +0.03 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.48 | 0.00 | +0.48 | + +## precision + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.41 | 0.64 | -0.23 | +| DeepSeek-V3-1 | 0.41 | 0.89 | -0.48 | +| GPT-4o-mini | 0.84 | 0.86 | -0.02 | +| GPT-5 | 0.84 | 0.91 | -0.07 | +| Gemini-2.5-flash | 0.57 | 0.81 | -0.23 | +| Gemini-2.5-flash-nothinking | 0.70 | 0.84 | -0.14 | +| Qwen3-235b | 0.78 | 0.85 | -0.07 | + +## recall + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.85 | 0.89 | -0.03 | +| DeepSeek-V3-1 | 0.84 | 0.90 | -0.05 | +| GPT-4o-mini | 0.98 | 0.92 | +0.07 | +| GPT-5 | 0.88 | 0.86 | +0.02 | +| Gemini-2.5-flash | 0.77 | 0.83 | -0.05 | +| Gemini-2.5-flash-nothinking | 0.78 | 0.80 | -0.02 | +| Qwen3-235b | 0.96 | 0.93 | +0.02 | + +## retry_rate + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.95 | 0.67 | +0.28 | +| DeepSeek-V3-1 | 1.00 | 0.27 | +0.73 | +| GPT-4o-mini | 0.17 | 0.28 | -0.12 | +| GPT-5 | 0.40 | 0.23 | +0.17 | +| Gemini-2.5-flash | 0.90 | 0.50 | +0.40 | +| Gemini-2.5-flash-nothinking | 0.75 | 0.37 | +0.38 | +| Qwen3-235b | 0.38 | 0.33 | +0.05 | + +## average_score + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.25 | 0.52 | -0.27 | +| DeepSeek-V3-1 | 0.12 | 0.53 | -0.41 | +| GPT-4o-mini | 0.58 | 0.58 | +0.00 | +| GPT-5 | 0.47 | 0.53 | -0.06 | +| Gemini-2.5-flash | 0.16 | 0.55 | -0.39 | +| Gemini-2.5-flash-nothinking | 0.05 | 0.57 | -0.52 | +| Qwen3-235b | 0.60 | 0.68 | -0.08 | + +# SocialMediaManager: A2A vs H_A2A + +## Overall Summary (Averaged Across All Models) + +| Metric | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| exact_match | 0.04 | 0.07 | -0.04 | +| any_order_match | 0.42 | 0.29 | +0.13 | +| precision | 0.80 | 0.85 | -0.05 | +| recall | 0.79 | 0.87 | -0.08 | +| retry_rate | 0.39 | 0.29 | +0.09 | +| pass_rate | 0.97 | 0.99 | -0.02 | + +--- + +## exact_match + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.00 | 0.02 | -0.02 | +| DeepSeek-V3-1 | 0.15 | 0.20 | -0.05 | +| GPT-4o-mini | 0.03 | 0.27 | -0.23 | +| GPT-5 | 0.05 | 0.00 | +0.05 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.03 | 0.03 | +0.00 | + +## any_order_match + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.13 | 0.32 | -0.18 | +| DeepSeek-V3-1 | 0.85 | 0.83 | +0.02 | +| GPT-4o-mini | 0.80 | 0.45 | +0.35 | +| GPT-5 | 0.53 | 0.15 | +0.38 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.62 | 0.27 | +0.35 | + +## precision + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.60 | 0.70 | -0.10 | +| DeepSeek-V3-1 | 0.79 | 0.83 | -0.04 | +| GPT-4o-mini | 0.66 | 0.90 | -0.24 | +| GPT-5 | 0.83 | 0.92 | -0.09 | +| Gemini-2.5-flash | 0.98 | 0.87 | +0.11 | +| Gemini-2.5-flash-nothinking | 0.97 | 0.93 | +0.04 | +| Qwen3-235b | 0.80 | 0.82 | -0.02 | + +## recall + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.89 | 0.90 | -0.01 | +| DeepSeek-V3-1 | 0.98 | 0.96 | +0.01 | +| GPT-4o-mini | 0.99 | 0.95 | +0.03 | +| GPT-5 | 0.90 | 0.87 | +0.02 | +| Gemini-2.5-flash | 0.44 | 0.77 | -0.33 | +| Gemini-2.5-flash-nothinking | 0.43 | 0.76 | -0.32 | +| Qwen3-235b | 0.93 | 0.88 | +0.05 | + +## retry_rate + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.65 | 0.55 | +0.10 | +| DeepSeek-V3-1 | 0.23 | 0.18 | +0.05 | +| GPT-4o-mini | 0.60 | 0.15 | +0.45 | +| GPT-5 | 0.48 | 0.08 | +0.40 | +| Gemini-2.5-flash | 0.07 | 0.42 | -0.35 | +| Gemini-2.5-flash-nothinking | 0.12 | 0.25 | -0.13 | +| Qwen3-235b | 0.55 | 0.42 | +0.13 | + +## pass_rate + +| Model | A2A | H_A2A | Diff (A2A-H_A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.92 | 0.97 | -0.05 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 0.93 | 1.00 | -0.07 | +| GPT-5 | 0.97 | 1.00 | -0.03 | +| Gemini-2.5-flash | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 1.00 | 1.00 | +0.00 | +| Qwen3-235b | 0.98 | 0.98 | +0.00 | + diff --git a/data/processed/RQ1/RadarCharts/Comparison_MCP_vs_A2A.md b/data/processed/RQ1/RadarCharts/Comparison_MCP_vs_A2A.md index f2fc018b85f006ff821914559a035b2b8fcd48e9..9bcd4d5ea39f51bdeb369e372b17c102caccf433 100644 --- a/data/processed/RQ1/RadarCharts/Comparison_MCP_vs_A2A.md +++ b/data/processed/RQ1/RadarCharts/Comparison_MCP_vs_A2A.md @@ -1,11 +1,18 @@ # Overall MCP vs A2A Comparison -Averaged across all projects: SQL_assistant, intelligent_recruitment_platform, landing_page_generator, self_evaluation_loop_flow, write_a_book_with_flows +Averaged across all projects: BookWriter, LandingPageGenerator, RecruitmentAssistant, SQLAssistant, SocialMediaManager ## Framework-Level Comparison (All Models Averaged) | Metric | MCP | A2A | Diff (MCP-A2A) | | --- | --- | --- | --- | +| exact_match | 0.07 | 0.01 | +0.06 | +| any_order_match | 0.29 | 0.25 | +0.04 | +| precision | 0.85 | 0.82 | +0.03 | +| recall | 0.81 | 0.84 | -0.03 | +| retry_rate | 0.21 | 0.22 | -0.02 | +| average_score | 0.47 | 0.31 | +0.17 | +| pass_rate | 0.99 | 0.99 | +0.00 | --- @@ -13,56 +20,520 @@ Averaged across all projects: SQL_assistant, intelligent_recruitment_platform, l | Model | MCP (Avg) | A2A (Avg) | Diff (MCP-A2A) | | --- | --- | --- | --- | +| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | +| DeepSeek-V3-1 | 0.20 | 0.03 | +0.17 | +| GPT-4o-mini | 0.04 | 0.01 | +0.03 | +| GPT-5 | 0.18 | 0.01 | +0.17 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.07 | 0.01 | +0.06 | ## any_order_match | Model | MCP (Avg) | A2A (Avg) | Diff (MCP-A2A) | | --- | --- | --- | --- | +| DeepSeek-R1 | 0.14 | 0.05 | +0.10 | +| DeepSeek-V3-1 | 0.63 | 0.48 | +0.15 | +| GPT-4o-mini | 0.54 | 0.60 | -0.06 | +| GPT-5 | 0.29 | 0.17 | +0.12 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.42 | 0.43 | -0.01 | ## precision | Model | MCP (Avg) | A2A (Avg) | Diff (MCP-A2A) | | --- | --- | --- | --- | +| DeepSeek-R1 | 0.71 | 0.70 | +0.02 | +| DeepSeek-V3-1 | 0.84 | 0.74 | +0.10 | +| GPT-4o-mini | 0.72 | 0.79 | -0.06 | +| GPT-5 | 0.90 | 0.89 | +0.02 | +| Gemini-2.5-flash | 0.97 | 0.91 | +0.06 | +| Gemini-2.5-flash-nothinking | 0.97 | 0.93 | +0.04 | +| Qwen3-235b | 0.84 | 0.81 | +0.03 | ## recall | Model | MCP (Avg) | A2A (Avg) | Diff (MCP-A2A) | | --- | --- | --- | --- | +| DeepSeek-R1 | 0.88 | 0.88 | -0.00 | +| DeepSeek-V3-1 | 0.98 | 0.95 | +0.02 | +| GPT-4o-mini | 0.95 | 0.96 | -0.01 | +| GPT-5 | 0.88 | 0.90 | -0.02 | +| Gemini-2.5-flash | 0.55 | 0.62 | -0.07 | +| Gemini-2.5-flash-nothinking | 0.57 | 0.64 | -0.07 | +| Qwen3-235b | 0.90 | 0.93 | -0.03 | ## retry_rate | Model | MCP (Avg) | A2A (Avg) | Diff (MCP-A2A) | | --- | --- | --- | --- | +| DeepSeek-R1 | 0.35 | 0.37 | -0.02 | +| DeepSeek-V3-1 | 0.19 | 0.25 | -0.06 | +| GPT-4o-mini | 0.26 | 0.15 | +0.11 | +| GPT-5 | 0.23 | 0.18 | +0.05 | +| Gemini-2.5-flash | 0.09 | 0.19 | -0.10 | +| Gemini-2.5-flash-nothinking | 0.09 | 0.18 | -0.09 | +| Qwen3-235b | 0.23 | 0.25 | -0.02 | ## average_score | Model | MCP (Avg) | A2A (Avg) | Diff (MCP-A2A) | | --- | --- | --- | --- | +| DeepSeek-R1 | 0.51 | 0.16 | +0.34 | +| DeepSeek-V3-1 | 0.66 | 0.49 | +0.17 | +| GPT-4o-mini | 0.63 | 0.57 | +0.06 | +| GPT-5 | 0.63 | 0.33 | +0.30 | +| Gemini-2.5-flash | 0.09 | 0.08 | +0.01 | +| Gemini-2.5-flash-nothinking | 0.10 | 0.03 | +0.08 | +| Qwen3-235b | 0.70 | 0.49 | +0.21 | ## pass_rate | Model | MCP (Avg) | A2A (Avg) | Diff (MCP-A2A) | | --- | --- | --- | --- | +| DeepSeek-R1 | 0.98 | 0.97 | +0.02 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 0.98 | 0.98 | +0.01 | +| GPT-5 | 0.99 | 0.99 | +0.01 | +| Gemini-2.5-flash | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 1.00 | 1.00 | +0.00 | +| Qwen3-235b | 0.98 | 0.99 | -0.01 | --- -# SQL_assistant: MCP vs A2A +# BookWriter: MCP vs A2A -_Data not available for comparison_ +## Overall Summary (Averaged Across All Models) -# intelligent_recruitment_platform: MCP vs A2A +| Metric | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| exact_match | 0.02 | 0.00 | +0.02 | +| any_order_match | 0.09 | 0.08 | +0.01 | +| precision | 0.94 | 0.93 | +0.01 | +| recall | 0.73 | 0.77 | -0.04 | +| retry_rate | 0.02 | 0.02 | +0.00 | +| pass_rate | 1.00 | 1.00 | +0.00 | + +--- + +## exact_match -_Data not available for comparison_ +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | +| DeepSeek-V3-1 | 0.13 | 0.00 | +0.13 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.00 | 0.00 | +0.00 | -# landing_page_generator: MCP vs A2A +## any_order_match -_Data not available for comparison_ +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.02 | 0.02 | +0.00 | +| DeepSeek-V3-1 | 0.63 | 0.53 | +0.10 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.00 | 0.00 | +0.00 | -# self_evaluation_loop_flow: MCP vs A2A +## precision -_Data not available for comparison_ +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.86 | 0.87 | -0.02 | +| DeepSeek-V3-1 | 0.96 | 0.79 | +0.17 | +| GPT-4o-mini | 0.84 | 0.90 | -0.06 | +| GPT-5 | 0.98 | 0.98 | +0.01 | +| Gemini-2.5-flash | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 1.00 | 1.00 | -0.00 | +| Qwen3-235b | 0.96 | 0.96 | -0.00 | + +## recall + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.80 | 0.84 | -0.04 | +| DeepSeek-V3-1 | 0.99 | 0.99 | -0.00 | +| GPT-4o-mini | 0.85 | 0.86 | -0.01 | +| GPT-5 | 0.81 | 0.82 | -0.02 | +| Gemini-2.5-flash | 0.47 | 0.53 | -0.06 | +| Gemini-2.5-flash-nothinking | 0.47 | 0.53 | -0.06 | +| Qwen3-235b | 0.73 | 0.80 | -0.06 | + +## retry_rate + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.13 | 0.13 | +0.00 | +| DeepSeek-V3-1 | 0.00 | 0.02 | -0.02 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.02 | 0.00 | +0.02 | +| Qwen3-235b | 0.02 | 0.00 | +0.02 | -# write_a_book_with_flows: MCP vs A2A +## pass_rate + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 1.00 | 1.00 | +0.00 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 1.00 | 1.00 | +0.00 | +| GPT-5 | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 1.00 | 1.00 | +0.00 | +| Qwen3-235b | 1.00 | 1.00 | +0.00 | + +# LandingPageGenerator: MCP vs A2A + +## Overall Summary (Averaged Across All Models) + +| Metric | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| exact_match | 0.29 | 0.00 | +0.29 | +| any_order_match | 0.60 | 0.40 | +0.20 | +| precision | 0.92 | 0.84 | +0.08 | +| recall | 0.90 | 0.91 | -0.01 | +| retry_rate | 0.03 | 0.02 | +0.01 | +| average_score | 0.57 | 0.30 | +0.27 | + +--- + +## exact_match + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | +| DeepSeek-V3-1 | 0.72 | 0.00 | +0.72 | +| GPT-4o-mini | 0.12 | 0.00 | +0.12 | +| GPT-5 | 0.88 | 0.00 | +0.88 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.28 | 0.00 | +0.28 | + +## any_order_match + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.32 | 0.05 | +0.27 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 0.98 | 0.97 | +0.02 | +| GPT-5 | 1.00 | 0.27 | +0.73 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.87 | 0.48 | +0.38 | -_Data not available for comparison_ +## precision + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.80 | 0.76 | +0.04 | +| DeepSeek-V3-1 | 0.97 | 0.83 | +0.13 | +| GPT-4o-mini | 0.82 | 0.74 | +0.08 | +| GPT-5 | 0.99 | 0.81 | +0.18 | +| Gemini-2.5-flash | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.99 | 0.98 | +0.00 | +| Qwen3-235b | 0.91 | 0.79 | +0.12 | + +## recall + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.94 | 0.94 | +0.00 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 1.00 | 1.00 | +0.00 | +| GPT-5 | 1.00 | 0.97 | +0.03 | +| Gemini-2.5-flash | 0.67 | 0.74 | -0.07 | +| Gemini-2.5-flash-nothinking | 0.70 | 0.76 | -0.07 | +| Qwen3-235b | 0.99 | 0.98 | +0.01 | + +## retry_rate + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.15 | 0.03 | +0.12 | +| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.03 | 0.05 | -0.02 | +| Qwen3-235b | 0.00 | 0.05 | -0.05 | + +## average_score + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.58 | 0.08 | +0.50 | +| DeepSeek-V3-1 | 0.83 | 0.86 | -0.02 | +| GPT-4o-mini | 0.86 | 0.57 | +0.30 | +| GPT-5 | 0.91 | 0.19 | +0.71 | +| Gemini-2.5-flash | 0.01 | 0.00 | +0.01 | +| Gemini-2.5-flash-nothinking | 0.03 | 0.00 | +0.03 | +| Qwen3-235b | 0.78 | 0.38 | +0.40 | + +# RecruitmentAssistant: MCP vs A2A + +## Overall Summary (Averaged Across All Models) + +| Metric | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| exact_match | 0.01 | 0.00 | +0.00 | +| any_order_match | 0.21 | 0.18 | +0.03 | +| precision | 0.87 | 0.89 | -0.02 | +| recall | 0.83 | 0.86 | -0.03 | +| retry_rate | 0.05 | 0.05 | +0.00 | +| pass_rate | 1.00 | 1.00 | +0.00 | + +--- + +## exact_match + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | +| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.05 | 0.02 | +0.03 | + +## any_order_match + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.08 | 0.00 | +0.08 | +| DeepSeek-V3-1 | 0.07 | 0.00 | +0.07 | +| GPT-4o-mini | 0.70 | 0.68 | +0.02 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.60 | 0.57 | +0.03 | + +## precision + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.79 | 0.83 | -0.04 | +| DeepSeek-V3-1 | 0.84 | 0.87 | -0.03 | +| GPT-4o-mini | 0.74 | 0.79 | -0.05 | +| GPT-5 | 0.99 | 0.99 | -0.00 | +| Gemini-2.5-flash | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.99 | 1.00 | -0.00 | +| Qwen3-235b | 0.73 | 0.74 | -0.01 | + +## recall + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.88 | 0.88 | +0.00 | +| DeepSeek-V3-1 | 0.94 | 0.96 | -0.02 | +| GPT-4o-mini | 0.98 | 0.98 | -0.01 | +| GPT-5 | 0.89 | 0.90 | -0.01 | +| Gemini-2.5-flash | 0.55 | 0.64 | -0.09 | +| Gemini-2.5-flash-nothinking | 0.60 | 0.68 | -0.08 | +| Qwen3-235b | 0.96 | 0.97 | -0.02 | + +## retry_rate + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.03 | 0.07 | -0.03 | +| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.33 | 0.27 | +0.07 | + +## pass_rate + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 1.00 | 0.98 | +0.02 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 1.00 | 1.00 | +0.00 | +| GPT-5 | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 1.00 | 1.00 | +0.00 | +| Qwen3-235b | 0.97 | 0.98 | -0.02 | + +# SQLAssistant: MCP vs A2A + +## Overall Summary (Averaged Across All Models) + +| Metric | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| exact_match | 0.00 | 0.00 | +0.00 | +| any_order_match | 0.14 | 0.15 | -0.02 | +| precision | 0.72 | 0.65 | +0.07 | +| recall | 0.86 | 0.87 | -0.01 | +| retry_rate | 0.53 | 0.65 | -0.12 | +| average_score | 0.38 | 0.32 | +0.06 | + +--- + +## exact_match + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | +| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | +| GPT-4o-mini | 0.00 | 0.00 | +0.00 | +| GPT-5 | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.00 | 0.00 | +0.00 | + +## any_order_match + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.10 | 0.03 | +0.07 | +| DeepSeek-V3-1 | 0.52 | 0.00 | +0.52 | +| GPT-4o-mini | 0.12 | 0.53 | -0.42 | +| GPT-5 | 0.00 | 0.03 | -0.03 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.23 | 0.48 | -0.25 | + +## precision + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.50 | 0.41 | +0.10 | +| DeepSeek-V3-1 | 0.67 | 0.41 | +0.26 | +| GPT-4o-mini | 0.61 | 0.84 | -0.23 | +| GPT-5 | 0.71 | 0.84 | -0.13 | +| Gemini-2.5-flash | 0.89 | 0.57 | +0.32 | +| Gemini-2.5-flash-nothinking | 0.90 | 0.70 | +0.20 | +| Qwen3-235b | 0.78 | 0.78 | -0.00 | + +## recall + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.90 | 0.85 | +0.05 | +| DeepSeek-V3-1 | 0.96 | 0.84 | +0.12 | +| GPT-4o-mini | 0.92 | 0.98 | -0.06 | +| GPT-5 | 0.85 | 0.88 | -0.04 | +| Gemini-2.5-flash | 0.71 | 0.77 | -0.07 | +| Gemini-2.5-flash-nothinking | 0.71 | 0.78 | -0.07 | +| Qwen3-235b | 0.96 | 0.96 | -0.00 | + +## retry_rate + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.78 | 0.95 | -0.17 | +| DeepSeek-V3-1 | 0.57 | 1.00 | -0.43 | +| GPT-4o-mini | 0.65 | 0.17 | +0.48 | +| GPT-5 | 0.68 | 0.40 | +0.28 | +| Gemini-2.5-flash | 0.30 | 0.90 | -0.60 | +| Gemini-2.5-flash-nothinking | 0.32 | 0.75 | -0.43 | +| Qwen3-235b | 0.38 | 0.38 | +0.00 | + +## average_score + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.44 | 0.25 | +0.19 | +| DeepSeek-V3-1 | 0.48 | 0.12 | +0.36 | +| GPT-4o-mini | 0.40 | 0.58 | -0.18 | +| GPT-5 | 0.36 | 0.47 | -0.11 | +| Gemini-2.5-flash | 0.16 | 0.16 | +0.01 | +| Gemini-2.5-flash-nothinking | 0.18 | 0.05 | +0.13 | +| Qwen3-235b | 0.62 | 0.60 | +0.02 | + +# SocialMediaManager: MCP vs A2A + +## Overall Summary (Averaged Across All Models) + +| Metric | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| exact_match | 0.03 | 0.04 | -0.01 | +| any_order_match | 0.40 | 0.42 | -0.01 | +| precision | 0.80 | 0.80 | -0.00 | +| recall | 0.75 | 0.79 | -0.04 | +| retry_rate | 0.40 | 0.39 | +0.01 | +| pass_rate | 0.98 | 0.97 | +0.01 | + +--- + +## exact_match + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | +| DeepSeek-V3-1 | 0.15 | 0.15 | +0.00 | +| GPT-4o-mini | 0.07 | 0.03 | +0.03 | +| GPT-5 | 0.00 | 0.05 | -0.05 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.00 | 0.03 | -0.03 | + +## any_order_match + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.20 | 0.13 | +0.07 | +| DeepSeek-V3-1 | 0.93 | 0.85 | +0.08 | +| GPT-4o-mini | 0.88 | 0.80 | +0.08 | +| GPT-5 | 0.43 | 0.53 | -0.10 | +| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | +| Qwen3-235b | 0.38 | 0.62 | -0.23 | + +## precision + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.62 | 0.60 | +0.02 | +| DeepSeek-V3-1 | 0.74 | 0.79 | -0.05 | +| GPT-4o-mini | 0.61 | 0.66 | -0.04 | +| GPT-5 | 0.85 | 0.83 | +0.02 | +| Gemini-2.5-flash | 0.97 | 0.98 | -0.02 | +| Gemini-2.5-flash-nothinking | 0.98 | 0.97 | +0.01 | +| Qwen3-235b | 0.85 | 0.80 | +0.05 | + +## recall + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.86 | 0.89 | -0.02 | +| DeepSeek-V3-1 | 0.98 | 0.98 | +0.01 | +| GPT-4o-mini | 0.99 | 0.99 | +0.00 | +| GPT-5 | 0.86 | 0.90 | -0.04 | +| Gemini-2.5-flash | 0.35 | 0.44 | -0.09 | +| Gemini-2.5-flash-nothinking | 0.35 | 0.43 | -0.09 | +| Qwen3-235b | 0.87 | 0.93 | -0.06 | + +## retry_rate + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.65 | 0.65 | +0.00 | +| DeepSeek-V3-1 | 0.38 | 0.23 | +0.15 | +| GPT-4o-mini | 0.67 | 0.60 | +0.07 | +| GPT-5 | 0.45 | 0.48 | -0.03 | +| Gemini-2.5-flash | 0.15 | 0.07 | +0.08 | +| Gemini-2.5-flash-nothinking | 0.08 | 0.12 | -0.03 | +| Qwen3-235b | 0.42 | 0.55 | -0.13 | + +## pass_rate + +| Model | MCP | A2A | Diff (MCP-A2A) | +| --- | --- | --- | --- | +| DeepSeek-R1 | 0.95 | 0.92 | +0.03 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 0.95 | 0.93 | +0.02 | +| GPT-5 | 0.98 | 0.97 | +0.02 | +| Gemini-2.5-flash | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 1.00 | 1.00 | +0.00 | +| Qwen3-235b | 0.98 | 0.98 | +0.00 | diff --git a/data/processed/RQ1/RadarCharts/Comparison_MCP_vs_Hardcoded.md b/data/processed/RQ1/RadarCharts/Comparison_MCP_vs_Hardcoded.md index e9f9120fda1d0e88d804b6ff0bdbd676e4a1b3df..a34ebd9e06a5cd865acb9af37b176067ff9f2733 100644 --- a/data/processed/RQ1/RadarCharts/Comparison_MCP_vs_Hardcoded.md +++ b/data/processed/RQ1/RadarCharts/Comparison_MCP_vs_Hardcoded.md @@ -1,6 +1,6 @@ # Overall MCP vs Hardcoded Comparison -Averaged across all projects: MarkdownValidator, GameBuilder, EmailResponder +Averaged across all projects: EmailResponder, GameBuilder, MarkdownValidator ## Framework-Level Comparison (All Models Averaged) @@ -12,7 +12,7 @@ Averaged across all projects: MarkdownValidator, GameBuilder, EmailResponder | recall | 0.91 | 0.91 | +0.00 | | retry_rate | 0.03 | 0.04 | -0.02 | | average_score | 0.60 | 0.60 | +0.00 | -| pass_rate | 0.00 | 0.00 | +0.00 | +| pass_rate | 1.00 | 0.99 | +0.01 | --- @@ -92,28 +92,28 @@ Averaged across all projects: MarkdownValidator, GameBuilder, EmailResponder | Model | MCP (Avg) | Hardcoded (Avg) | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | -| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | -| GPT-4o-mini | 0.00 | 0.00 | +0.00 | -| GPT-5 | 0.00 | 0.00 | +0.00 | -| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | -| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | -| Qwen3-235b | 0.00 | 0.00 | +0.00 | +| DeepSeek-R1 | 1.00 | 0.99 | +0.01 | +| DeepSeek-V3-1 | 1.00 | 0.98 | +0.02 | +| GPT-4o-mini | 1.00 | 1.00 | +0.00 | +| GPT-5 | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash | 1.00 | 0.99 | +0.01 | +| Gemini-2.5-flash-nothinking | 1.00 | 1.00 | +0.00 | +| Qwen3-235b | 1.00 | 1.00 | +0.00 | --- -# MarkdownValidator: MCP vs Hardcoded +# EmailResponder: MCP vs Hardcoded ## Overall Summary (Averaged Across All Models) | Metric | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| exact_match | 0.75 | 0.77 | -0.02 | -| any_order_match | 0.76 | 0.79 | -0.03 | -| precision | 0.98 | 0.98 | -0.00 | -| recall | 0.94 | 0.95 | -0.00 | -| retry_rate | 0.04 | 0.04 | +0.00 | -| average_score | 0.60 | 0.60 | +0.00 | +| exact_match | 0.10 | 0.10 | -0.01 | +| any_order_match | 0.14 | 0.11 | +0.02 | +| precision | 0.96 | 0.95 | +0.01 | +| recall | 0.85 | 0.84 | +0.01 | +| retry_rate | 0.04 | 0.07 | -0.04 | +| pass_rate | 1.00 | 1.00 | +0.00 | --- @@ -121,73 +121,73 @@ Averaged across all projects: MarkdownValidator, GameBuilder, EmailResponder | Model | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| DeepSeek-R1 | 0.97 | 1.00 | -0.03 | -| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | -| GPT-4o-mini | 0.98 | 0.97 | +0.02 | -| GPT-5 | 1.00 | 1.00 | +0.00 | -| Gemini-2.5-flash | 0.20 | 0.32 | -0.12 | -| Gemini-2.5-flash-nothinking | 0.10 | 0.17 | -0.07 | -| Qwen3-235b | 1.00 | 0.97 | +0.03 | +| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | +| DeepSeek-V3-1 | 0.06 | 0.26 | -0.20 | +| GPT-4o-mini | 0.56 | 0.30 | +0.26 | +| GPT-5 | 0.00 | 0.01 | -0.01 | +| Gemini-2.5-flash | 0.00 | 0.01 | -0.01 | +| Gemini-2.5-flash-nothinking | 0.02 | 0.07 | -0.05 | +| Qwen3-235b | 0.04 | 0.08 | -0.04 | ## any_order_match | Model | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| DeepSeek-R1 | 1.00 | 1.00 | +0.00 | -| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | -| GPT-4o-mini | 1.00 | 1.00 | +0.00 | -| GPT-5 | 1.00 | 1.00 | +0.00 | -| Gemini-2.5-flash | 0.25 | 0.40 | -0.15 | -| Gemini-2.5-flash-nothinking | 0.10 | 0.17 | -0.07 | -| Qwen3-235b | 1.00 | 0.97 | +0.03 | +| DeepSeek-R1 | 0.03 | 0.02 | +0.01 | +| DeepSeek-V3-1 | 0.22 | 0.26 | -0.04 | +| GPT-4o-mini | 0.61 | 0.34 | +0.27 | +| GPT-5 | 0.00 | 0.01 | -0.01 | +| Gemini-2.5-flash | 0.00 | 0.01 | -0.01 | +| Gemini-2.5-flash-nothinking | 0.05 | 0.07 | -0.02 | +| Qwen3-235b | 0.05 | 0.09 | -0.04 | ## precision | Model | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| DeepSeek-R1 | 0.99 | 1.00 | -0.01 | -| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | -| GPT-4o-mini | 1.00 | 0.99 | +0.00 | +| DeepSeek-R1 | 0.80 | 0.74 | +0.06 | +| DeepSeek-V3-1 | 0.97 | 0.99 | -0.03 | +| GPT-4o-mini | 0.99 | 0.99 | +0.00 | | GPT-5 | 1.00 | 1.00 | +0.00 | -| Gemini-2.5-flash | 0.88 | 0.86 | +0.02 | -| Gemini-2.5-flash-nothinking | 0.95 | 0.97 | -0.02 | -| Qwen3-235b | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash | 0.97 | 0.97 | +0.01 | +| Gemini-2.5-flash-nothinking | 0.96 | 0.96 | +0.00 | +| Qwen3-235b | 1.00 | 1.00 | -0.00 | ## recall | Model | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| DeepSeek-R1 | 1.00 | 1.00 | +0.00 | -| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | -| GPT-4o-mini | 1.00 | 1.00 | +0.00 | -| GPT-5 | 1.00 | 1.00 | +0.00 | -| Gemini-2.5-flash | 0.86 | 0.89 | -0.03 | -| Gemini-2.5-flash-nothinking | 0.73 | 0.74 | -0.01 | -| Qwen3-235b | 1.00 | 0.99 | +0.01 | +| DeepSeek-R1 | 0.90 | 0.89 | +0.00 | +| DeepSeek-V3-1 | 0.88 | 0.88 | -0.00 | +| GPT-4o-mini | 0.94 | 0.89 | +0.04 | +| GPT-5 | 0.83 | 0.83 | -0.00 | +| Gemini-2.5-flash | 0.75 | 0.72 | +0.04 | +| Gemini-2.5-flash-nothinking | 0.80 | 0.82 | -0.01 | +| Qwen3-235b | 0.84 | 0.85 | -0.00 | ## retry_rate | Model | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | -| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | +| DeepSeek-R1 | 0.04 | 0.28 | -0.24 | +| DeepSeek-V3-1 | 0.02 | 0.02 | +0.00 | | GPT-4o-mini | 0.00 | 0.00 | +0.00 | | GPT-5 | 0.00 | 0.00 | +0.00 | -| Gemini-2.5-flash | 0.20 | 0.22 | -0.02 | -| Gemini-2.5-flash-nothinking | 0.07 | 0.05 | +0.02 | +| Gemini-2.5-flash | 0.09 | 0.10 | -0.01 | +| Gemini-2.5-flash-nothinking | 0.12 | 0.12 | +0.00 | | Qwen3-235b | 0.00 | 0.00 | +0.00 | -## average_score +## pass_rate | Model | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| DeepSeek-R1 | 0.79 | 0.82 | -0.03 | -| DeepSeek-V3-1 | 0.84 | 0.85 | -0.01 | -| GPT-4o-mini | 0.84 | 0.81 | +0.03 | -| GPT-5 | 0.70 | 0.72 | -0.03 | -| Gemini-2.5-flash | 0.06 | 0.11 | -0.05 | -| Gemini-2.5-flash-nothinking | 0.18 | 0.10 | +0.08 | -| Qwen3-235b | 0.83 | 0.78 | +0.04 | +| DeepSeek-R1 | 1.00 | 0.98 | +0.02 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 1.00 | 1.00 | +0.00 | +| GPT-5 | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash | 1.00 | 0.99 | +0.01 | +| Gemini-2.5-flash-nothinking | 1.00 | 1.00 | +0.00 | +| Qwen3-235b | 1.00 | 1.00 | +0.00 | # GameBuilder: MCP vs Hardcoded @@ -200,7 +200,7 @@ Averaged across all projects: MarkdownValidator, GameBuilder, EmailResponder | precision | 0.98 | 0.99 | -0.01 | | recall | 0.94 | 0.93 | +0.00 | | retry_rate | 0.00 | 0.01 | -0.01 | -| pass_rate | 0.00 | 0.00 | +0.00 | +| pass_rate | 1.00 | 0.99 | +0.01 | --- @@ -268,26 +268,26 @@ Averaged across all projects: MarkdownValidator, GameBuilder, EmailResponder | Model | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | -| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | -| GPT-4o-mini | 0.00 | 0.00 | +0.00 | -| GPT-5 | 0.00 | 0.00 | +0.00 | -| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | -| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | -| Qwen3-235b | 0.00 | 0.00 | +0.00 | +| DeepSeek-R1 | 1.00 | 1.00 | +0.00 | +| DeepSeek-V3-1 | 1.00 | 0.96 | +0.04 | +| GPT-4o-mini | 1.00 | 1.00 | +0.00 | +| GPT-5 | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash-nothinking | 1.00 | 1.00 | +0.00 | +| Qwen3-235b | 1.00 | 1.00 | +0.00 | -# EmailResponder: MCP vs Hardcoded +# MarkdownValidator: MCP vs Hardcoded ## Overall Summary (Averaged Across All Models) | Metric | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| exact_match | 0.10 | 0.10 | -0.01 | -| any_order_match | 0.14 | 0.11 | +0.02 | -| precision | 0.96 | 0.95 | +0.01 | -| recall | 0.85 | 0.84 | +0.01 | -| retry_rate | 0.04 | 0.07 | -0.04 | -| pass_rate | 0.00 | 0.00 | +0.00 | +| exact_match | 0.75 | 0.77 | -0.02 | +| any_order_match | 0.76 | 0.79 | -0.03 | +| precision | 0.98 | 0.98 | -0.00 | +| recall | 0.94 | 0.95 | -0.00 | +| retry_rate | 0.04 | 0.04 | +0.00 | +| average_score | 0.60 | 0.60 | +0.00 | --- @@ -295,71 +295,71 @@ Averaged across all projects: MarkdownValidator, GameBuilder, EmailResponder | Model | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | -| DeepSeek-V3-1 | 0.06 | 0.26 | -0.20 | -| GPT-4o-mini | 0.56 | 0.30 | +0.26 | -| GPT-5 | 0.00 | 0.01 | -0.01 | -| Gemini-2.5-flash | 0.00 | 0.01 | -0.01 | -| Gemini-2.5-flash-nothinking | 0.02 | 0.07 | -0.05 | -| Qwen3-235b | 0.04 | 0.08 | -0.04 | +| DeepSeek-R1 | 0.97 | 1.00 | -0.03 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 0.98 | 0.97 | +0.02 | +| GPT-5 | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash | 0.20 | 0.32 | -0.12 | +| Gemini-2.5-flash-nothinking | 0.10 | 0.17 | -0.07 | +| Qwen3-235b | 1.00 | 0.97 | +0.03 | ## any_order_match | Model | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| DeepSeek-R1 | 0.03 | 0.02 | +0.01 | -| DeepSeek-V3-1 | 0.22 | 0.26 | -0.04 | -| GPT-4o-mini | 0.61 | 0.34 | +0.27 | -| GPT-5 | 0.00 | 0.01 | -0.01 | -| Gemini-2.5-flash | 0.00 | 0.01 | -0.01 | -| Gemini-2.5-flash-nothinking | 0.05 | 0.07 | -0.02 | -| Qwen3-235b | 0.05 | 0.09 | -0.04 | +| DeepSeek-R1 | 1.00 | 1.00 | +0.00 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 1.00 | 1.00 | +0.00 | +| GPT-5 | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash | 0.25 | 0.40 | -0.15 | +| Gemini-2.5-flash-nothinking | 0.10 | 0.17 | -0.07 | +| Qwen3-235b | 1.00 | 0.97 | +0.03 | ## precision | Model | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| DeepSeek-R1 | 0.80 | 0.74 | +0.06 | -| DeepSeek-V3-1 | 0.97 | 0.99 | -0.03 | -| GPT-4o-mini | 0.99 | 0.99 | +0.00 | +| DeepSeek-R1 | 0.99 | 1.00 | -0.01 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 1.00 | 0.99 | +0.00 | | GPT-5 | 1.00 | 1.00 | +0.00 | -| Gemini-2.5-flash | 0.97 | 0.97 | +0.01 | -| Gemini-2.5-flash-nothinking | 0.96 | 0.96 | +0.00 | -| Qwen3-235b | 1.00 | 1.00 | -0.00 | +| Gemini-2.5-flash | 0.88 | 0.86 | +0.02 | +| Gemini-2.5-flash-nothinking | 0.95 | 0.97 | -0.02 | +| Qwen3-235b | 1.00 | 1.00 | +0.00 | ## recall | Model | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| DeepSeek-R1 | 0.90 | 0.89 | +0.00 | -| DeepSeek-V3-1 | 0.88 | 0.88 | -0.00 | -| GPT-4o-mini | 0.94 | 0.89 | +0.04 | -| GPT-5 | 0.83 | 0.83 | -0.00 | -| Gemini-2.5-flash | 0.75 | 0.72 | +0.04 | -| Gemini-2.5-flash-nothinking | 0.80 | 0.82 | -0.01 | -| Qwen3-235b | 0.84 | 0.85 | -0.00 | +| DeepSeek-R1 | 1.00 | 1.00 | +0.00 | +| DeepSeek-V3-1 | 1.00 | 1.00 | +0.00 | +| GPT-4o-mini | 1.00 | 1.00 | +0.00 | +| GPT-5 | 1.00 | 1.00 | +0.00 | +| Gemini-2.5-flash | 0.86 | 0.89 | -0.03 | +| Gemini-2.5-flash-nothinking | 0.73 | 0.74 | -0.01 | +| Qwen3-235b | 1.00 | 0.99 | +0.01 | ## retry_rate | Model | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| DeepSeek-R1 | 0.04 | 0.28 | -0.24 | -| DeepSeek-V3-1 | 0.02 | 0.02 | +0.00 | +| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | +| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | | GPT-4o-mini | 0.00 | 0.00 | +0.00 | | GPT-5 | 0.00 | 0.00 | +0.00 | -| Gemini-2.5-flash | 0.09 | 0.10 | -0.01 | -| Gemini-2.5-flash-nothinking | 0.12 | 0.12 | +0.00 | +| Gemini-2.5-flash | 0.20 | 0.22 | -0.02 | +| Gemini-2.5-flash-nothinking | 0.07 | 0.05 | +0.02 | | Qwen3-235b | 0.00 | 0.00 | +0.00 | -## pass_rate +## average_score | Model | MCP | Hardcoded | Diff (MCP-Hardcoded) | | --- | --- | --- | --- | -| DeepSeek-R1 | 0.00 | 0.00 | +0.00 | -| DeepSeek-V3-1 | 0.00 | 0.00 | +0.00 | -| GPT-4o-mini | 0.00 | 0.00 | +0.00 | -| GPT-5 | 0.00 | 0.00 | +0.00 | -| Gemini-2.5-flash | 0.00 | 0.00 | +0.00 | -| Gemini-2.5-flash-nothinking | 0.00 | 0.00 | +0.00 | -| Qwen3-235b | 0.00 | 0.00 | +0.00 | +| DeepSeek-R1 | 0.79 | 0.82 | -0.03 | +| DeepSeek-V3-1 | 0.84 | 0.85 | -0.01 | +| GPT-4o-mini | 0.84 | 0.81 | +0.03 | +| GPT-5 | 0.70 | 0.72 | -0.03 | +| Gemini-2.5-flash | 0.06 | 0.11 | -0.05 | +| Gemini-2.5-flash-nothinking | 0.18 | 0.10 | +0.08 | +| Qwen3-235b | 0.83 | 0.78 | +0.04 | diff --git a/data/processed/RQ1/RadarCharts/Comparison_Powerful_vs_Lightweight.md b/data/processed/RQ1/RadarCharts/Comparison_Powerful_vs_Lightweight.md index e2074f990c78e30474a2c863d3493ba92eca4c8e..578efa73f0a6bd3cf9d40d79e43c312efd0e2e03 100644 --- a/data/processed/RQ1/RadarCharts/Comparison_Powerful_vs_Lightweight.md +++ b/data/processed/RQ1/RadarCharts/Comparison_Powerful_vs_Lightweight.md @@ -2,15 +2,17 @@ ## Overall Summary (Averaged Across All Project Series) +_Note: average_score is averaged only across GT projects; pass_rate is averaged only across non-GT projects._ + | Metric | GPT-5 | GPT-4o-mini | Diff (A-B) | | --- | --- | --- | --- | -| exact_match | 0.19 | 0.20 | -0.01 | -| any_order_match | 0.25 | 0.58 | -0.32 | -| precision | 0.92 | 0.83 | +0.09 | -| recall | 0.89 | 0.95 | -0.06 | -| retry_rate | 0.10 | 0.11 | -0.01 | -| average_score | 0.56 | 0.68 | -0.12 | -| pass_rate | 0.68 | 0.71 | -0.03 | +| exact_match | 0.29 | 0.31 | -0.02 | +| any_order_match | 0.35 | 0.64 | -0.28 | +| precision | 0.94 | 0.85 | +0.09 | +| recall | 0.90 | 0.95 | -0.05 | +| retry_rate | 0.10 | 0.10 | -0.01 | +| average_score | 0.55 | 0.67 | -0.13 | +| pass_rate | 1.00 | 0.99 | +0.00 | --- @@ -19,111 +21,81 @@ | Project Series | GPT-5 | GPT-4o-mini | Diff (A-B) | | --- | --- | --- | --- | | BookWriter | 0.00 | 0.00 | +0.00 | -| BookWriter-H_A2A | 0.00 | 0.00 | +0.00 | -| EmailResponder | 0.01 | 0.43 | -0.43 | +| EmailResponder | 0.01 | 0.43 | -0.42 | | GameBuilder | 1.00 | 0.88 | +0.12 | -| LandingPageGenerator | 0.44 | 0.06 | +0.38 | -| LandingPageGenerator-H_A2A | 0.00 | 0.00 | +0.00 | +| LandingPageGenerator | 0.29 | 0.04 | +0.26 | | MarkdownValidator | 1.00 | 0.97 | +0.03 | | RecruitmentAssistant | 0.00 | 0.00 | +0.00 | -| RecruitmentAssistant-H_A2A | 0.00 | 0.00 | +0.00 | | SQLAssistant | 0.00 | 0.00 | +0.00 | -| SQLAssistant-H_A2A | 0.00 | 0.00 | +0.00 | -| SocialMediaManager | 0.03 | 0.05 | -0.02 | -| SocialMediaManager-H_A2A | 0.00 | 0.27 | -0.27 | +| SocialMediaManager | 0.02 | 0.12 | -0.11 | ## any_order_match | Project Series | GPT-5 | GPT-4o-mini | Diff (A-B) | | --- | --- | --- | --- | | BookWriter | 0.00 | 0.00 | +0.00 | -| BookWriter-H_A2A | 0.00 | 0.00 | +0.00 | | EmailResponder | 0.01 | 0.47 | -0.47 | | GameBuilder | 1.00 | 1.00 | +0.00 | -| LandingPageGenerator | 0.63 | 0.97 | -0.34 | -| LandingPageGenerator-H_A2A | 0.00 | 1.00 | -1.00 | +| LandingPageGenerator | 0.42 | 0.98 | -0.56 | | MarkdownValidator | 1.00 | 1.00 | +0.00 | -| RecruitmentAssistant | 0.00 | 0.69 | -0.69 | -| RecruitmentAssistant-H_A2A | 0.00 | 0.72 | -0.72 | -| SQLAssistant | 0.02 | 0.32 | -0.31 | -| SQLAssistant-H_A2A | 0.00 | 0.00 | +0.00 | -| SocialMediaManager | 0.48 | 0.84 | -0.36 | -| SocialMediaManager-H_A2A | 0.15 | 0.45 | -0.30 | +| RecruitmentAssistant | 0.00 | 0.70 | -0.70 | +| SQLAssistant | 0.01 | 0.22 | -0.21 | +| SocialMediaManager | 0.37 | 0.71 | -0.34 | ## precision | Project Series | GPT-5 | GPT-4o-mini | Diff (A-B) | | --- | --- | --- | --- | -| BookWriter | 0.98 | 0.87 | +0.11 | -| BookWriter-H_A2A | 0.95 | 0.80 | +0.15 | +| BookWriter | 0.97 | 0.85 | +0.12 | | EmailResponder | 1.00 | 0.99 | +0.01 | | GameBuilder | 1.00 | 0.96 | +0.04 | -| LandingPageGenerator | 0.90 | 0.78 | +0.12 | -| LandingPageGenerator-H_A2A | 0.79 | 0.78 | +0.02 | +| LandingPageGenerator | 0.86 | 0.78 | +0.08 | | MarkdownValidator | 1.00 | 0.99 | +0.01 | -| RecruitmentAssistant | 0.99 | 0.77 | +0.22 | -| RecruitmentAssistant-H_A2A | 0.97 | 0.75 | +0.22 | -| SQLAssistant | 0.77 | 0.73 | +0.05 | -| SQLAssistant-H_A2A | 0.91 | 0.86 | +0.05 | -| SocialMediaManager | 0.84 | 0.63 | +0.20 | -| SocialMediaManager-H_A2A | 0.92 | 0.90 | +0.02 | +| RecruitmentAssistant | 0.98 | 0.76 | +0.22 | +| SQLAssistant | 0.82 | 0.77 | +0.05 | +| SocialMediaManager | 0.86 | 0.72 | +0.14 | ## recall | Project Series | GPT-5 | GPT-4o-mini | Diff (A-B) | | --- | --- | --- | --- | -| BookWriter | 0.81 | 0.86 | -0.04 | -| BookWriter-H_A2A | 0.67 | 0.77 | -0.10 | +| BookWriter | 0.77 | 0.83 | -0.06 | | EmailResponder | 0.83 | 0.91 | -0.08 | | GameBuilder | 1.00 | 1.00 | +0.00 | -| LandingPageGenerator | 0.99 | 1.00 | -0.01 | -| LandingPageGenerator-H_A2A | 0.94 | 1.00 | -0.06 | +| LandingPageGenerator | 0.97 | 1.00 | -0.03 | | MarkdownValidator | 1.00 | 1.00 | +0.00 | | RecruitmentAssistant | 0.90 | 0.98 | -0.08 | -| RecruitmentAssistant-H_A2A | 0.91 | 0.99 | -0.08 | -| SQLAssistant | 0.87 | 0.95 | -0.09 | -| SQLAssistant-H_A2A | 0.86 | 0.92 | -0.05 | -| SocialMediaManager | 0.88 | 0.99 | -0.11 | -| SocialMediaManager-H_A2A | 0.87 | 0.95 | -0.08 | +| SQLAssistant | 0.86 | 0.94 | -0.08 | +| SocialMediaManager | 0.88 | 0.97 | -0.10 | ## retry_rate | Project Series | GPT-5 | GPT-4o-mini | Diff (A-B) | | --- | --- | --- | --- | | BookWriter | 0.00 | 0.00 | +0.00 | -| BookWriter-H_A2A | 0.00 | 0.00 | +0.00 | | EmailResponder | 0.00 | 0.00 | +0.00 | | GameBuilder | 0.00 | 0.00 | +0.00 | | LandingPageGenerator | 0.00 | 0.00 | +0.00 | -| LandingPageGenerator-H_A2A | 0.00 | 0.00 | +0.00 | | MarkdownValidator | 0.00 | 0.00 | +0.00 | | RecruitmentAssistant | 0.00 | 0.00 | +0.00 | -| RecruitmentAssistant-H_A2A | 0.00 | 0.00 | +0.00 | -| SQLAssistant | 0.54 | 0.41 | +0.13 | -| SQLAssistant-H_A2A | 0.23 | 0.28 | -0.05 | -| SocialMediaManager | 0.47 | 0.63 | -0.17 | -| SocialMediaManager-H_A2A | 0.08 | 0.15 | -0.07 | +| SQLAssistant | 0.44 | 0.37 | +0.07 | +| SocialMediaManager | 0.34 | 0.47 | -0.13 | ## average_score | Project Series | GPT-5 | GPT-4o-mini | Diff (A-B) | | --- | --- | --- | --- | -| LandingPageGenerator | 0.55 | 0.72 | -0.17 | +| LandingPageGenerator | 0.47 | 0.67 | -0.20 | | MarkdownValidator | 0.71 | 0.82 | -0.11 | -| SQLAssistant | 0.41 | 0.49 | -0.08 | +| SQLAssistant | 0.45 | 0.52 | -0.07 | ## pass_rate | Project Series | GPT-5 | GPT-4o-mini | Diff (A-B) | | --- | --- | --- | --- | | BookWriter | 1.00 | 1.00 | +0.00 | -| BookWriter-H_A2A | 1.00 | 1.00 | +0.00 | -| EmailResponder | 0.00 | 0.00 | +0.00 | -| GameBuilder | 0.00 | 0.00 | +0.00 | -| LandingPageGenerator-H_A2A | 0.32 | 0.59 | -0.27 | +| EmailResponder | 1.00 | 1.00 | +0.00 | +| GameBuilder | 1.00 | 1.00 | +0.00 | | RecruitmentAssistant | 1.00 | 1.00 | +0.00 | -| RecruitmentAssistant-H_A2A | 1.00 | 1.00 | +0.00 | -| SQLAssistant-H_A2A | 0.53 | 0.58 | -0.05 | -| SocialMediaManager | 0.97 | 0.94 | +0.03 | -| SocialMediaManager-H_A2A | 1.00 | 1.00 | +0.00 | +| SocialMediaManager | 0.98 | 0.96 | +0.02 | diff --git a/data/processed/RQ1/RadarCharts/Comparison_Reasoning_vs_NonReasoning.md b/data/processed/RQ1/RadarCharts/Comparison_Reasoning_vs_NonReasoning.md index 10c0beb26584fbc1e778852de78a22a2d9207591..b76b39758ee4f37fdfc3f75dcf8975c0acdf4140 100644 --- a/data/processed/RQ1/RadarCharts/Comparison_Reasoning_vs_NonReasoning.md +++ b/data/processed/RQ1/RadarCharts/Comparison_Reasoning_vs_NonReasoning.md @@ -2,15 +2,17 @@ ## Overall Summary (Averaged Across All Project Series) +_Note: average_score is averaged only across GT projects; pass_rate is averaged only across non-GT projects._ + | Metric | DeepSeek-R1 | DeepSeek-V3-1 | Diff (A-B) | | --- | --- | --- | --- | -| exact_match | 0.15 | 0.22 | -0.07 | -| any_order_match | 0.24 | 0.65 | -0.41 | -| precision | 0.79 | 0.85 | -0.07 | -| recall | 0.90 | 0.96 | -0.06 | -| retry_rate | 0.25 | 0.12 | +0.13 | -| average_score | 0.49 | 0.66 | -0.17 | -| pass_rate | 0.65 | 0.74 | -0.09 | +| exact_match | 0.24 | 0.31 | -0.07 | +| any_order_match | 0.31 | 0.65 | -0.34 | +| precision | 0.80 | 0.87 | -0.07 | +| recall | 0.91 | 0.96 | -0.05 | +| retry_rate | 0.22 | 0.11 | +0.11 | +| average_score | 0.49 | 0.69 | -0.20 | +| pass_rate | 0.99 | 1.00 | -0.01 | --- @@ -18,114 +20,84 @@ | Project Series | DeepSeek-R1 | DeepSeek-V3-1 | Diff (A-B) | | --- | --- | --- | --- | -| BookWriter | 0.00 | 0.07 | -0.07 | -| BookWriter-H_A2A | 0.00 | 0.00 | +0.00 | +| BookWriter | 0.00 | 0.04 | -0.04 | | EmailResponder | 0.00 | 0.16 | -0.16 | | GameBuilder | 0.94 | 0.90 | +0.04 | -| LandingPageGenerator | 0.00 | 0.36 | -0.36 | -| LandingPageGenerator-H_A2A | 0.00 | 0.00 | +0.00 | +| LandingPageGenerator | 0.00 | 0.24 | -0.24 | | MarkdownValidator | 0.98 | 1.00 | -0.02 | | RecruitmentAssistant | 0.00 | 0.00 | +0.00 | -| RecruitmentAssistant-H_A2A | 0.00 | 0.00 | +0.00 | | SQLAssistant | 0.00 | 0.00 | +0.00 | -| SQLAssistant-H_A2A | 0.00 | 0.00 | +0.00 | -| SocialMediaManager | 0.00 | 0.15 | -0.15 | -| SocialMediaManager-H_A2A | 0.02 | 0.20 | -0.18 | +| SocialMediaManager | 0.01 | 0.17 | -0.16 | ## any_order_match | Project Series | DeepSeek-R1 | DeepSeek-V3-1 | Diff (A-B) | | --- | --- | --- | --- | -| BookWriter | 0.02 | 0.58 | -0.57 | -| BookWriter-H_A2A | 0.00 | 0.58 | -0.58 | +| BookWriter | 0.01 | 0.58 | -0.57 | | EmailResponder | 0.03 | 0.24 | -0.21 | | GameBuilder | 0.96 | 1.00 | -0.04 | -| LandingPageGenerator | 0.18 | 1.00 | -0.82 | -| LandingPageGenerator-H_A2A | 0.03 | 0.98 | -0.95 | +| LandingPageGenerator | 0.13 | 0.99 | -0.86 | | MarkdownValidator | 1.00 | 1.00 | +0.00 | -| RecruitmentAssistant | 0.04 | 0.03 | +0.01 | -| RecruitmentAssistant-H_A2A | 0.25 | 1.00 | -0.75 | -| SQLAssistant | 0.07 | 0.26 | -0.19 | -| SQLAssistant-H_A2A | 0.00 | 0.00 | +0.00 | -| SocialMediaManager | 0.17 | 0.89 | -0.72 | -| SocialMediaManager-H_A2A | 0.32 | 0.83 | -0.52 | +| RecruitmentAssistant | 0.11 | 0.36 | -0.24 | +| SQLAssistant | 0.04 | 0.17 | -0.13 | +| SocialMediaManager | 0.22 | 0.87 | -0.66 | ## precision | Project Series | DeepSeek-R1 | DeepSeek-V3-1 | Diff (A-B) | | --- | --- | --- | --- | -| BookWriter | 0.87 | 0.88 | -0.01 | -| BookWriter-H_A2A | 0.91 | 0.92 | -0.01 | +| BookWriter | 0.88 | 0.89 | -0.01 | | EmailResponder | 0.77 | 0.98 | -0.21 | | GameBuilder | 1.00 | 0.97 | +0.03 | -| LandingPageGenerator | 0.78 | 0.90 | -0.12 | -| LandingPageGenerator-H_A2A | 0.78 | 0.77 | +0.01 | +| LandingPageGenerator | 0.78 | 0.86 | -0.08 | | MarkdownValidator | 1.00 | 1.00 | -0.00 | -| RecruitmentAssistant | 0.81 | 0.85 | -0.04 | -| RecruitmentAssistant-H_A2A | 0.89 | 0.82 | +0.08 | -| SQLAssistant | 0.46 | 0.54 | -0.09 | -| SQLAssistant-H_A2A | 0.64 | 0.89 | -0.25 | -| SocialMediaManager | 0.61 | 0.76 | -0.15 | -| SocialMediaManager-H_A2A | 0.70 | 0.83 | -0.13 | +| RecruitmentAssistant | 0.84 | 0.84 | -0.00 | +| SQLAssistant | 0.52 | 0.66 | -0.14 | +| SocialMediaManager | 0.64 | 0.78 | -0.14 | ## recall | Project Series | DeepSeek-R1 | DeepSeek-V3-1 | Diff (A-B) | | --- | --- | --- | --- | -| BookWriter | 0.82 | 0.99 | -0.17 | -| BookWriter-H_A2A | 0.77 | 0.95 | -0.18 | +| BookWriter | 0.81 | 0.98 | -0.17 | | EmailResponder | 0.89 | 0.88 | +0.02 | | GameBuilder | 0.99 | 1.00 | -0.01 | | LandingPageGenerator | 0.94 | 1.00 | -0.06 | -| LandingPageGenerator-H_A2A | 0.93 | 1.00 | -0.07 | | MarkdownValidator | 1.00 | 1.00 | +0.00 | -| RecruitmentAssistant | 0.88 | 0.95 | -0.07 | -| RecruitmentAssistant-H_A2A | 0.93 | 1.00 | -0.07 | -| SQLAssistant | 0.88 | 0.90 | -0.03 | -| SQLAssistant-H_A2A | 0.89 | 0.90 | -0.01 | -| SocialMediaManager | 0.88 | 0.98 | -0.10 | -| SocialMediaManager-H_A2A | 0.90 | 0.96 | -0.06 | +| RecruitmentAssistant | 0.90 | 0.97 | -0.07 | +| SQLAssistant | 0.88 | 0.90 | -0.02 | +| SocialMediaManager | 0.88 | 0.97 | -0.09 | ## retry_rate | Project Series | DeepSeek-R1 | DeepSeek-V3-1 | Diff (A-B) | | --- | --- | --- | --- | -| BookWriter | 0.13 | 0.01 | +0.12 | -| BookWriter-H_A2A | 0.02 | 0.00 | +0.02 | +| BookWriter | 0.09 | 0.01 | +0.09 | | EmailResponder | 0.16 | 0.02 | +0.14 | | GameBuilder | 0.00 | 0.00 | +0.00 | -| LandingPageGenerator | 0.09 | 0.00 | +0.09 | -| LandingPageGenerator-H_A2A | 0.00 | 0.03 | -0.03 | +| LandingPageGenerator | 0.06 | 0.01 | +0.05 | | MarkdownValidator | 0.00 | 0.00 | +0.00 | | RecruitmentAssistant | 0.05 | 0.00 | +0.05 | -| RecruitmentAssistant-H_A2A | 0.05 | 0.00 | +0.05 | -| SQLAssistant | 0.87 | 0.78 | +0.08 | -| SQLAssistant-H_A2A | 0.67 | 0.27 | +0.40 | -| SocialMediaManager | 0.65 | 0.31 | +0.34 | -| SocialMediaManager-H_A2A | 0.55 | 0.18 | +0.37 | +| SQLAssistant | 0.80 | 0.61 | +0.19 | +| SocialMediaManager | 0.62 | 0.27 | +0.35 | ## average_score | Project Series | DeepSeek-R1 | DeepSeek-V3-1 | Diff (A-B) | | --- | --- | --- | --- | -| LandingPageGenerator | 0.33 | 0.85 | -0.52 | +| LandingPageGenerator | 0.25 | 0.86 | -0.60 | | MarkdownValidator | 0.81 | 0.84 | -0.04 | -| SQLAssistant | 0.34 | 0.30 | +0.04 | +| SQLAssistant | 0.40 | 0.37 | +0.03 | ## pass_rate | Project Series | DeepSeek-R1 | DeepSeek-V3-1 | Diff (A-B) | | --- | --- | --- | --- | | BookWriter | 1.00 | 1.00 | +0.00 | -| BookWriter-H_A2A | 1.00 | 1.00 | +0.00 | -| EmailResponder | 0.00 | 0.00 | +0.00 | -| GameBuilder | 0.00 | 0.00 | +0.00 | -| LandingPageGenerator-H_A2A | 0.10 | 0.88 | -0.77 | +| EmailResponder | 0.99 | 1.00 | -0.01 | +| GameBuilder | 1.00 | 0.98 | +0.02 | | RecruitmentAssistant | 0.99 | 1.00 | -0.01 | -| RecruitmentAssistant-H_A2A | 1.00 | 1.00 | +0.00 | -| SQLAssistant-H_A2A | 0.52 | 0.53 | -0.01 | -| SocialMediaManager | 0.93 | 1.00 | -0.07 | -| SocialMediaManager-H_A2A | 0.97 | 1.00 | -0.03 | +| SocialMediaManager | 0.94 | 1.00 | -0.06 | --- @@ -134,15 +106,17 @@ ## Overall Summary (Averaged Across All Project Series) +_Note: average_score is averaged only across GT projects; pass_rate is averaged only across non-GT projects._ + | Metric | Gemini-2.5-flash | Gemini-2.5-flash-nothinking | Diff (A-B) | | --- | --- | --- | --- | -| exact_match | 0.02 | 0.02 | +0.00 | -| any_order_match | 0.03 | 0.02 | +0.01 | -| precision | 0.92 | 0.94 | -0.01 | -| recall | 0.71 | 0.71 | -0.00 | -| retry_rate | 0.15 | 0.12 | +0.03 | -| average_score | 0.08 | 0.09 | -0.01 | -| pass_rate | 0.65 | 0.66 | -0.00 | +| exact_match | 0.03 | 0.02 | +0.01 | +| any_order_match | 0.04 | 0.03 | +0.01 | +| precision | 0.93 | 0.95 | -0.01 | +| recall | 0.70 | 0.71 | -0.00 | +| retry_rate | 0.14 | 0.11 | +0.03 | +| average_score | 0.13 | 0.14 | -0.01 | +| pass_rate | 1.00 | 1.00 | -0.00 | --- @@ -151,111 +125,81 @@ | Project Series | Gemini-2.5-flash | Gemini-2.5-flash-nothinking | Diff (A-B) | | --- | --- | --- | --- | | BookWriter | 0.00 | 0.00 | +0.00 | -| BookWriter-H_A2A | 0.00 | 0.00 | +0.00 | -| EmailResponder | 0.01 | 0.05 | -0.04 | +| EmailResponder | 0.01 | 0.04 | -0.04 | | GameBuilder | 0.00 | 0.02 | -0.02 | | LandingPageGenerator | 0.00 | 0.00 | +0.00 | -| LandingPageGenerator-H_A2A | 0.00 | 0.00 | +0.00 | | MarkdownValidator | 0.26 | 0.13 | +0.12 | | RecruitmentAssistant | 0.00 | 0.00 | +0.00 | -| RecruitmentAssistant-H_A2A | 0.00 | 0.00 | +0.00 | | SQLAssistant | 0.00 | 0.00 | +0.00 | -| SQLAssistant-H_A2A | 0.00 | 0.00 | +0.00 | | SocialMediaManager | 0.00 | 0.00 | +0.00 | -| SocialMediaManager-H_A2A | 0.00 | 0.00 | +0.00 | ## any_order_match | Project Series | Gemini-2.5-flash | Gemini-2.5-flash-nothinking | Diff (A-B) | | --- | --- | --- | --- | | BookWriter | 0.00 | 0.00 | +0.00 | -| BookWriter-H_A2A | 0.00 | 0.00 | +0.00 | | EmailResponder | 0.01 | 0.06 | -0.06 | | GameBuilder | 0.00 | 0.06 | -0.06 | | LandingPageGenerator | 0.00 | 0.00 | +0.00 | -| LandingPageGenerator-H_A2A | 0.00 | 0.00 | +0.00 | | MarkdownValidator | 0.33 | 0.13 | +0.19 | | RecruitmentAssistant | 0.00 | 0.00 | +0.00 | -| RecruitmentAssistant-H_A2A | 0.00 | 0.00 | +0.00 | | SQLAssistant | 0.00 | 0.00 | +0.00 | -| SQLAssistant-H_A2A | 0.00 | 0.00 | +0.00 | | SocialMediaManager | 0.00 | 0.00 | +0.00 | -| SocialMediaManager-H_A2A | 0.00 | 0.00 | +0.00 | ## precision | Project Series | Gemini-2.5-flash | Gemini-2.5-flash-nothinking | Diff (A-B) | | --- | --- | --- | --- | -| BookWriter | 1.00 | 1.00 | +0.00 | -| BookWriter-H_A2A | 0.93 | 0.93 | -0.00 | +| BookWriter | 0.98 | 0.98 | +0.00 | | EmailResponder | 0.97 | 0.96 | +0.01 | | GameBuilder | 1.00 | 0.99 | +0.01 | -| LandingPageGenerator | 1.00 | 0.99 | +0.01 | -| LandingPageGenerator-H_A2A | 0.91 | 0.87 | +0.05 | +| LandingPageGenerator | 0.97 | 0.95 | +0.02 | | MarkdownValidator | 0.87 | 0.96 | -0.09 | -| RecruitmentAssistant | 1.00 | 0.99 | +0.01 | -| RecruitmentAssistant-H_A2A | 0.94 | 0.93 | +0.01 | -| SQLAssistant | 0.73 | 0.80 | -0.07 | -| SQLAssistant-H_A2A | 0.81 | 0.84 | -0.03 | -| SocialMediaManager | 0.97 | 0.98 | -0.00 | -| SocialMediaManager-H_A2A | 0.87 | 0.93 | -0.05 | +| RecruitmentAssistant | 0.98 | 0.97 | +0.01 | +| SQLAssistant | 0.76 | 0.82 | -0.06 | +| SocialMediaManager | 0.94 | 0.96 | -0.02 | ## recall | Project Series | Gemini-2.5-flash | Gemini-2.5-flash-nothinking | Diff (A-B) | | --- | --- | --- | --- | -| BookWriter | 0.50 | 0.50 | -0.01 | -| BookWriter-H_A2A | 0.57 | 0.57 | -0.00 | +| BookWriter | 0.52 | 0.53 | -0.01 | | EmailResponder | 0.74 | 0.81 | -0.07 | | GameBuilder | 0.75 | 0.80 | -0.04 | -| LandingPageGenerator | 0.70 | 0.73 | -0.03 | -| LandingPageGenerator-H_A2A | 0.87 | 0.85 | +0.02 | +| LandingPageGenerator | 0.76 | 0.77 | -0.01 | | MarkdownValidator | 0.88 | 0.73 | +0.15 | -| RecruitmentAssistant | 0.59 | 0.64 | -0.05 | -| RecruitmentAssistant-H_A2A | 0.89 | 0.94 | -0.05 | -| SQLAssistant | 0.74 | 0.74 | -0.00 | -| SQLAssistant-H_A2A | 0.83 | 0.80 | +0.03 | -| SocialMediaManager | 0.39 | 0.39 | +0.00 | -| SocialMediaManager-H_A2A | 0.77 | 0.76 | +0.01 | +| RecruitmentAssistant | 0.69 | 0.74 | -0.05 | +| SQLAssistant | 0.77 | 0.76 | +0.01 | +| SocialMediaManager | 0.52 | 0.51 | +0.01 | ## retry_rate | Project Series | Gemini-2.5-flash | Gemini-2.5-flash-nothinking | Diff (A-B) | | --- | --- | --- | --- | | BookWriter | 0.00 | 0.01 | -0.01 | -| BookWriter-H_A2A | 0.00 | 0.00 | +0.00 | | EmailResponder | 0.10 | 0.12 | -0.02 | | GameBuilder | 0.00 | 0.04 | -0.04 | -| LandingPageGenerator | 0.00 | 0.04 | -0.04 | -| LandingPageGenerator-H_A2A | 0.00 | 0.00 | +0.00 | +| LandingPageGenerator | 0.00 | 0.03 | -0.03 | | MarkdownValidator | 0.21 | 0.06 | +0.15 | | RecruitmentAssistant | 0.00 | 0.00 | +0.00 | -| RecruitmentAssistant-H_A2A | 0.00 | 0.00 | +0.00 | -| SQLAssistant | 0.60 | 0.53 | +0.07 | -| SQLAssistant-H_A2A | 0.50 | 0.37 | +0.13 | -| SocialMediaManager | 0.11 | 0.10 | +0.01 | -| SocialMediaManager-H_A2A | 0.42 | 0.25 | +0.17 | +| SQLAssistant | 0.57 | 0.48 | +0.09 | +| SocialMediaManager | 0.21 | 0.15 | +0.06 | ## average_score | Project Series | Gemini-2.5-flash | Gemini-2.5-flash-nothinking | Diff (A-B) | | --- | --- | --- | --- | -| LandingPageGenerator | 0.01 | 0.01 | -0.01 | +| LandingPageGenerator | 0.00 | 0.01 | -0.00 | | MarkdownValidator | 0.08 | 0.14 | -0.06 | -| SQLAssistant | 0.16 | 0.12 | +0.04 | +| SQLAssistant | 0.29 | 0.27 | +0.02 | ## pass_rate | Project Series | Gemini-2.5-flash | Gemini-2.5-flash-nothinking | Diff (A-B) | | --- | --- | --- | --- | | BookWriter | 1.00 | 1.00 | +0.00 | -| BookWriter-H_A2A | 1.00 | 1.00 | +0.00 | -| EmailResponder | 0.00 | 0.00 | +0.00 | -| GameBuilder | 0.00 | 0.00 | +0.00 | -| LandingPageGenerator-H_A2A | 0.00 | 0.00 | +0.00 | +| EmailResponder | 0.99 | 1.00 | -0.01 | +| GameBuilder | 1.00 | 1.00 | +0.00 | | RecruitmentAssistant | 1.00 | 1.00 | +0.00 | -| RecruitmentAssistant-H_A2A | 1.00 | 1.00 | +0.00 | -| SQLAssistant-H_A2A | 0.55 | 0.57 | -0.02 | | SocialMediaManager | 1.00 | 1.00 | +0.00 | -| SocialMediaManager-H_A2A | 1.00 | 1.00 | +0.00 | diff --git a/data/processed/RQ1/RadarCharts/EmailResponder-MCP_radar.pdf b/data/processed/RQ1/RadarCharts/EmailResponder-MCP_radar.pdf index eb8e2965cc5173b912807ecfa0bb582aa53f774a..f8512b7bf6a622edc786d721abde7dc2857cd622 100644 --- a/data/processed/RQ1/RadarCharts/EmailResponder-MCP_radar.pdf +++ b/data/processed/RQ1/RadarCharts/EmailResponder-MCP_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:54d373300d71e6253fe1b8406ce65bfe890ba34f2e4451aded082b77890ca09a -size 18382 +oid sha256:82319b0db8093476b91a21f85c27bed9819c7f7c12f806a6a28563d07f45c67a +size 18722 diff --git a/data/processed/RQ1/RadarCharts/EmailResponder_Series_Average_radar.pdf b/data/processed/RQ1/RadarCharts/EmailResponder_Series_Average_radar.pdf index 579a934a6a794f4108a597ae4c6d887ee94cfe7a..292977e34603d070a9d02e1ea14461300a1b4941 100644 --- a/data/processed/RQ1/RadarCharts/EmailResponder_Series_Average_radar.pdf +++ b/data/processed/RQ1/RadarCharts/EmailResponder_Series_Average_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:51a8bf69d906d1e2ab9bd501c4963bdc5becebea100cfda1bc29fed5973e6fc8 -size 18507 +oid sha256:f560590ef5841252a5eb343e6accc9d341e4e435f57c0332bee8d44c4cbecfe2 +size 18641 diff --git a/data/processed/RQ1/RadarCharts/EmailResponder_radar.pdf b/data/processed/RQ1/RadarCharts/EmailResponder_radar.pdf index 2bf1b24d3484255167757ba46f98563eb70324ac..33f645817213dc9574b50ac5d1e13dc407229e8a 100644 --- a/data/processed/RQ1/RadarCharts/EmailResponder_radar.pdf +++ b/data/processed/RQ1/RadarCharts/EmailResponder_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:49e9ad9ae4ec5cac5b46163d0cac8081bc4971e74af0ddbf66fb1c55e17db977 -size 18469 +oid sha256:8a3d796f43b54ad5e5be1edd2fb1ea3082dff038e57af8391579ca04c4545198 +size 18806 diff --git a/data/processed/RQ1/RadarCharts/GameBuilder-MCP_radar.pdf b/data/processed/RQ1/RadarCharts/GameBuilder-MCP_radar.pdf index 8fb5f7a2f84bced52ed812ea0d543fb37235c8a5..81557a658626c5f86814d733935aca32210b01a7 100644 --- a/data/processed/RQ1/RadarCharts/GameBuilder-MCP_radar.pdf +++ b/data/processed/RQ1/RadarCharts/GameBuilder-MCP_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:b06beed8b0114b670733470116ea3db358369c95170c5df462c80fcc1642358c -size 17899 +oid sha256:250cb641161793638113d3b4be23a941428cb3815d034c33bec6288df8d27889 +size 18147 diff --git a/data/processed/RQ1/RadarCharts/GameBuilder_Series_Average_radar.pdf b/data/processed/RQ1/RadarCharts/GameBuilder_Series_Average_radar.pdf index 71225dd5bc24b6da8e042d281d833926da382cf5..e8513a1b7e62231a9b64031c68a744b39f8f62d9 100644 --- a/data/processed/RQ1/RadarCharts/GameBuilder_Series_Average_radar.pdf +++ b/data/processed/RQ1/RadarCharts/GameBuilder_Series_Average_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:819850734e505a15eef48dd8f98ac48b0bded688c5ba03386d2d2bbd17775dd9 -size 18193 +oid sha256:70a2da74ccf57f910114d86a35342513fec68e6c0c8549ad7d7dfa6926e97c49 +size 18252 diff --git a/data/processed/RQ1/RadarCharts/GameBuilder_radar.pdf b/data/processed/RQ1/RadarCharts/GameBuilder_radar.pdf index e3d54f2f8d4e5954a1d5021fa4af207ef725a50b..89d15712d5a2c407cedac65ea02e0e78bed3d255 100644 --- a/data/processed/RQ1/RadarCharts/GameBuilder_radar.pdf +++ b/data/processed/RQ1/RadarCharts/GameBuilder_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:79d67995b5802b1d4ae17a733a63265a9b54b2e0450f3976f0f0a73539175445 -size 18121 +oid sha256:f3f06c27070a9b069d10231f785d1a512030d3351106d0bed6f7576f8ef9606b +size 18417 diff --git a/data/processed/RQ1/RadarCharts/LandingPageGenerator-A2A_radar.pdf b/data/processed/RQ1/RadarCharts/LandingPageGenerator-A2A_radar.pdf index 419774f8f732f5a40b80d2f894c6ee51da8c6e61..76ad960ca776f1805ea83a8c44ce74b97c129e04 100644 --- a/data/processed/RQ1/RadarCharts/LandingPageGenerator-A2A_radar.pdf +++ b/data/processed/RQ1/RadarCharts/LandingPageGenerator-A2A_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:cded14c20b009b59de83839a071af3ca7a2da52e14ff7ecba7e8c5eb637602a9 -size 19474 +oid sha256:6e283a4e3b66898455d53a5641cdd8a7f468032e825e7d6677410dba6712e63a +size 19572 diff --git a/data/processed/RQ1/RadarCharts/LandingPageGenerator-H_A2A_radar.pdf b/data/processed/RQ1/RadarCharts/LandingPageGenerator-H_A2A_radar.pdf index be1474a1a2bdd3ecca2b965ef1bb220524582a4b..5acafc7f4671ac4fdbf54d3d24fa005a2f0855eb 100644 --- a/data/processed/RQ1/RadarCharts/LandingPageGenerator-H_A2A_radar.pdf +++ b/data/processed/RQ1/RadarCharts/LandingPageGenerator-H_A2A_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:a313052eb1966fa4719914ab50906368ddb3bbdf074cda1a7211ecdd3c65c2d6 -size 19399 +oid sha256:606eeda4856ee1e6b396aee9f0850eb48b0b5267f2d7af99b77ea800f3df7737 +size 19484 diff --git a/data/processed/RQ1/RadarCharts/LandingPageGenerator-MCP_radar.pdf b/data/processed/RQ1/RadarCharts/LandingPageGenerator-MCP_radar.pdf index 0a15bfb25660422ad4526fa1e4c73bef89e7e7c2..c673fc8dbea73f69bb657c681a9ea293ee744469 100644 --- a/data/processed/RQ1/RadarCharts/LandingPageGenerator-MCP_radar.pdf +++ b/data/processed/RQ1/RadarCharts/LandingPageGenerator-MCP_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:8620b8f80a74059960fd37dbd14091b07a35cd9e05f925224df71e8926ae3c0c -size 19590 +oid sha256:c45d54c583452d556450852f197f5b2e4a72f44a7de346cac4696b4384dbd66a +size 19594 diff --git a/data/processed/RQ1/RadarCharts/LandingPageGenerator_Series_Average_radar.pdf b/data/processed/RQ1/RadarCharts/LandingPageGenerator_Series_Average_radar.pdf index f5792ecf50f6127e8d72c907b2a02dbf13c20295..1abe2d236392986adcb449bd5ed49dd5088b5135 100644 --- a/data/processed/RQ1/RadarCharts/LandingPageGenerator_Series_Average_radar.pdf +++ b/data/processed/RQ1/RadarCharts/LandingPageGenerator_Series_Average_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:9a4809a49812fab174e384f7127ebb674632e88af91a1a539037560d87e728de -size 19789 +oid sha256:39e9f915b575c973593c41cf04078e3fa61bb370917238ab079e86c95fcf5019 +size 19739 diff --git a/data/processed/RQ1/RadarCharts/MarkdownValidator-MCP_radar.pdf b/data/processed/RQ1/RadarCharts/MarkdownValidator-MCP_radar.pdf index a4a64cb9390bfd3dcb9879095595adfed3586f5d..94a32b8b982992de62c6524eaa4490be1dbd9223 100644 --- a/data/processed/RQ1/RadarCharts/MarkdownValidator-MCP_radar.pdf +++ b/data/processed/RQ1/RadarCharts/MarkdownValidator-MCP_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:eb66c55b36236874cf1f86f4ca99a26dd1142092d5cc9f7ba84b351a39bd8692 -size 19351 +oid sha256:85ab29d260711210659302dd5d67d8335dc7e93168fa6282811e7e4a4ec4f987 +size 19187 diff --git a/data/processed/RQ1/RadarCharts/MarkdownValidator_Series_Average_radar.pdf b/data/processed/RQ1/RadarCharts/MarkdownValidator_Series_Average_radar.pdf index 8b09b09bab41cd04b3ffaa7cf85c355c7e06216c..c66339ecd9d13bb18d58a1bf2bbfa829965bd302 100644 --- a/data/processed/RQ1/RadarCharts/MarkdownValidator_Series_Average_radar.pdf +++ b/data/processed/RQ1/RadarCharts/MarkdownValidator_Series_Average_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:28c2d81f35928dfa2dd2fd89f003e5d50360e03249704f07db250865fdf24e11 -size 19441 +oid sha256:1b6c2486f67c77a03b45609e427f58afc1bb7c63f08de8d47985d40c6c0d250e +size 19115 diff --git a/data/processed/RQ1/RadarCharts/MarkdownValidator_radar.pdf b/data/processed/RQ1/RadarCharts/MarkdownValidator_radar.pdf index a9b621be264647d10030d1a9ffec064869d555b2..c2fdb67f2103722d548c1c52e90936f8ff896b06 100644 --- a/data/processed/RQ1/RadarCharts/MarkdownValidator_radar.pdf +++ b/data/processed/RQ1/RadarCharts/MarkdownValidator_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:54351eaa5fe68c9a18c899d5f81b584b73f82c920020a5f151ecedc7601036b2 -size 19357 +oid sha256:f950d188d24702bf10707a12104307d334ad8c1bfcf7ac9fec4d7d1fa3c2c6e6 +size 19195 diff --git a/data/processed/RQ1/RadarCharts/Model_Legend.pdf b/data/processed/RQ1/RadarCharts/Model_Legend.pdf index c7f662ad3251daac9b2623481f0bf22fe377ef96..f69045b680b3149502d57d67a783d2833b26ae35 100644 --- a/data/processed/RQ1/RadarCharts/Model_Legend.pdf +++ b/data/processed/RQ1/RadarCharts/Model_Legend.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:87995800ac65dd63355b90ca0c2b435f4938cb47a4f64248c171ca64c2b7283e +oid sha256:83288982959a945c33345d775bbd7ec017865674864eb226ced527df37f6f7c8 size 16883 diff --git a/data/processed/RQ1/RadarCharts/Model_Legend_horizontal.pdf b/data/processed/RQ1/RadarCharts/Model_Legend_horizontal.pdf index 0a348ca416b6ebc82f8718f50be823e33713631a..42f0f05ea51a68fbd6850472f24b964086d280e7 100644 --- a/data/processed/RQ1/RadarCharts/Model_Legend_horizontal.pdf +++ b/data/processed/RQ1/RadarCharts/Model_Legend_horizontal.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:85eff9bd2173f5f51feb68a7c8c641f40a40cbf341093a50176ee8ae7ae051ea +oid sha256:a0833597bb1b7972025c1d714c10b3c8e7ad4e511061d1dc67e5d24d15a8486c size 16715 diff --git a/data/processed/RQ1/RadarCharts/Overall_Average_radar.pdf b/data/processed/RQ1/RadarCharts/Overall_Average_radar.pdf index d61422e279412d88fe22d95ddf373a662633e6fe..336aa8596c5fa629be852b86d2796cceeca53a44 100644 --- a/data/processed/RQ1/RadarCharts/Overall_Average_radar.pdf +++ b/data/processed/RQ1/RadarCharts/Overall_Average_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:d9a05de9b2aea9728fc9b3c311398a2f1f1032686972e005ca6bbf655cc733da -size 19089 +oid sha256:45566e18abbcba3bdecb82bb9474de2410ccdf5402fa8c30e4c158a488e16283 +size 19086 diff --git a/data/processed/RQ1/RadarCharts/RecruitmentAssistant-A2A_radar.pdf b/data/processed/RQ1/RadarCharts/RecruitmentAssistant-A2A_radar.pdf index dcf65b641b5583e3f723317bdafe8c105f1db71d..1318176ed592a3e71d4a5494e4af4ec0473022c5 100644 --- a/data/processed/RQ1/RadarCharts/RecruitmentAssistant-A2A_radar.pdf +++ b/data/processed/RQ1/RadarCharts/RecruitmentAssistant-A2A_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:4d2f7e94106a1e2acab6cbdbfac95f29e8a9eb1c3ce185017a0243217edd7287 -size 18188 +oid sha256:df38763b3d0257faea18cae23f51dc8926b446fb14d9d5050866f10f59187f4a +size 18443 diff --git a/data/processed/RQ1/RadarCharts/RecruitmentAssistant-H_A2A_radar.pdf b/data/processed/RQ1/RadarCharts/RecruitmentAssistant-H_A2A_radar.pdf index e3c45f1ca4fd07841f3abc62af2688a683fafb37..93d2eb6367b36b92a122e6116a7395738de9df14 100644 --- a/data/processed/RQ1/RadarCharts/RecruitmentAssistant-H_A2A_radar.pdf +++ b/data/processed/RQ1/RadarCharts/RecruitmentAssistant-H_A2A_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:4a234d1cca8e73d55dedf0126199c3b59597b90dd541f02cae457634ec8feb32 -size 18250 +oid sha256:4cceac688835bc2f7afd6d10bc151023d4bf159b4a28b4f4507d11496bb8bced +size 18534 diff --git a/data/processed/RQ1/RadarCharts/RecruitmentAssistant-MCP_radar.pdf b/data/processed/RQ1/RadarCharts/RecruitmentAssistant-MCP_radar.pdf index a026fd7543b2fd407372471eda34b003e8e15b21..116d5e441c10f987ac48d67b123cb661ce5b4f13 100644 --- a/data/processed/RQ1/RadarCharts/RecruitmentAssistant-MCP_radar.pdf +++ b/data/processed/RQ1/RadarCharts/RecruitmentAssistant-MCP_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:67d2260a99fc6a67dfe3bf7305d246f1b8863e27a3aea65d5b4c2e3094e27e07 -size 18286 +oid sha256:c76ccaf0d68501098cf7152cf0d39b8819acbd1dbddf58e662838cf8a413202c +size 18533 diff --git a/data/processed/RQ1/RadarCharts/RecruitmentAssistant_Series_Average_radar.pdf b/data/processed/RQ1/RadarCharts/RecruitmentAssistant_Series_Average_radar.pdf index e578d6d02300b690b4bb0cbd7fe07d648fd2cc2c..79d6d6524361d6bffc280594579b239e5ab777e9 100644 --- a/data/processed/RQ1/RadarCharts/RecruitmentAssistant_Series_Average_radar.pdf +++ b/data/processed/RQ1/RadarCharts/RecruitmentAssistant_Series_Average_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:01f7047bdc69d6ee35741573c4bb9b0abeefef4777d1dbf99858a344390a58a8 +oid sha256:3fa043db14b497da71d4b0b5b8aa8e5fb0019c429b938cfe5a628bcaa4056319 size 18359 diff --git a/data/processed/RQ1/RadarCharts/SQLAssistant-A2A_radar.pdf b/data/processed/RQ1/RadarCharts/SQLAssistant-A2A_radar.pdf index 09d4b0b0bb601555476f5659e60c5063d1f3f3e8..41de1208abfe40b67b3afc3ec9763ecedd61d42d 100644 --- a/data/processed/RQ1/RadarCharts/SQLAssistant-A2A_radar.pdf +++ b/data/processed/RQ1/RadarCharts/SQLAssistant-A2A_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:dde9c1fb231791991b05329b9e68b62ed93c574921d209472f8d46a4eaba117f -size 19696 +oid sha256:3f960fe9b801913e81635335c9f6a3e6dab60a3db4fa07c7cfc126f114298ba2 +size 19699 diff --git a/data/processed/RQ1/RadarCharts/SQLAssistant-H_A2A_radar.pdf b/data/processed/RQ1/RadarCharts/SQLAssistant-H_A2A_radar.pdf index 5e40bba9e41d7348e5e498840b67410297c40156..399cf0b764547ac311ed9eabc7bd09ced92673fd 100644 --- a/data/processed/RQ1/RadarCharts/SQLAssistant-H_A2A_radar.pdf +++ b/data/processed/RQ1/RadarCharts/SQLAssistant-H_A2A_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:13028d9a20022c3cbfd116fa7acb377ffc6919aff5b088d8436027517d2278d1 +oid sha256:0b3c9afbdabbcee68732350554888d7f9cbe20e4269ac8282b2dbdd143540934 size 19561 diff --git a/data/processed/RQ1/RadarCharts/SQLAssistant-MCP_radar.pdf b/data/processed/RQ1/RadarCharts/SQLAssistant-MCP_radar.pdf index 548be7f1be20afc95b80bfcb822a2171de087278..b144fb83bbd8f42d54adda11d7a67551b2e73973 100644 --- a/data/processed/RQ1/RadarCharts/SQLAssistant-MCP_radar.pdf +++ b/data/processed/RQ1/RadarCharts/SQLAssistant-MCP_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:26d35e21a146e98e50d5a5b7c84ff5ee5c7e6494238a60215641b74d488f7896 -size 19779 +oid sha256:84329640c5fa3fa4ac9d071e223b02fe5bb44d12c423340278cb619afcac9eb6 +size 19774 diff --git a/data/processed/RQ1/RadarCharts/SQLAssistant_Series_Average_radar.pdf b/data/processed/RQ1/RadarCharts/SQLAssistant_Series_Average_radar.pdf index d29ae8e63d4e9a58998b24208e70b60ea843443b..0fe2822b678ba67cffb3f9c5a1890fbd20265080 100644 --- a/data/processed/RQ1/RadarCharts/SQLAssistant_Series_Average_radar.pdf +++ b/data/processed/RQ1/RadarCharts/SQLAssistant_Series_Average_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:ce8114a22bc99d936c92ebb9f2b296f9705abb9c7b0e1706a2d539450b5de86e -size 19812 +oid sha256:78aed82dbd503751c823ca9f450da06e20f21655891197e4db1b011d943853cb +size 19655 diff --git a/data/processed/RQ1/RadarCharts/SocialMediaManager-A2A_radar.pdf b/data/processed/RQ1/RadarCharts/SocialMediaManager-A2A_radar.pdf index a1c8e82050d20e8608a60ab56c82db358009108c..96b492f3d322ea72d8a694f616f4aae464ba2ed4 100644 --- a/data/processed/RQ1/RadarCharts/SocialMediaManager-A2A_radar.pdf +++ b/data/processed/RQ1/RadarCharts/SocialMediaManager-A2A_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:3b9cfffcb0634fc47adfe8389af02d687dcc77143251c907ce0670484a401b9e -size 18799 +oid sha256:a6d5ebbbb3b7a8891ca9452a6f0965d045d00bac44f9cc23043beee43787cd68 +size 18896 diff --git a/data/processed/RQ1/RadarCharts/SocialMediaManager-H_A2A_radar.pdf b/data/processed/RQ1/RadarCharts/SocialMediaManager-H_A2A_radar.pdf index 7e1856d112edc2eeb72c7837d928f7f75364be2b..40bbbd264598c9d3bbc7a790a98206898a0ab34b 100644 --- a/data/processed/RQ1/RadarCharts/SocialMediaManager-H_A2A_radar.pdf +++ b/data/processed/RQ1/RadarCharts/SocialMediaManager-H_A2A_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:5f02145d11843714ba35b535a92bda34ef9503ee3ca79436a69c8e616a5ceb00 -size 18703 +oid sha256:164a922c634b36c13aea9a4ed57cf5fe93e2bf021d160231bae766547965fa53 +size 18870 diff --git a/data/processed/RQ1/RadarCharts/SocialMediaManager-MCP_radar.pdf b/data/processed/RQ1/RadarCharts/SocialMediaManager-MCP_radar.pdf index 175720306307938d5870ed17eeebe62e2c278eee..4262e91044972e604d36f3d53beacd8032d4e9bc 100644 --- a/data/processed/RQ1/RadarCharts/SocialMediaManager-MCP_radar.pdf +++ b/data/processed/RQ1/RadarCharts/SocialMediaManager-MCP_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:e847cabc6bb9b85966068252272e3eb08a0a0259fbfb1f2c46e6d993c90148aa -size 18695 +oid sha256:7e231b761b29b9e7a11cbdf747a5103b740413c3d625d7dec34427730ddef935 +size 18870 diff --git a/data/processed/RQ1/RadarCharts/SocialMediaManager_Series_Average_radar.pdf b/data/processed/RQ1/RadarCharts/SocialMediaManager_Series_Average_radar.pdf index 70f8b3e3a44202384a74a99324745a4613d88781..7b7e829f22593da2af898e25b66d6145fb9f9040 100644 --- a/data/processed/RQ1/RadarCharts/SocialMediaManager_Series_Average_radar.pdf +++ b/data/processed/RQ1/RadarCharts/SocialMediaManager_Series_Average_radar.pdf @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:a97c6dd871fa3a86e28af835d736d50106b29917465b7c5fb2ffda6fff8810e0 +oid sha256:7323b2edb827896dca89a963c9ed4561511319fb288725cefc25925bfa60df92 size 18835 diff --git a/data/processed/RQ1/__pycache__/generate_radar_charts.cpython-313.pyc b/data/processed/RQ1/__pycache__/generate_radar_charts.cpython-313.pyc new file mode 100644 index 0000000000000000000000000000000000000000..6e45df9755884f46b10187a825be81a64f38050f Binary files /dev/null and b/data/processed/RQ1/__pycache__/generate_radar_charts.cpython-313.pyc differ diff --git a/data/processed/RQ1/generate_radar_charts.py b/data/processed/RQ1/generate_radar_charts.py index e7a24b49d1bc74919891741a67a399819e2aa156..7baf414e5791185d5466112db17f6fef389fd07b 100644 --- a/data/processed/RQ1/generate_radar_charts.py +++ b/data/processed/RQ1/generate_radar_charts.py @@ -9,6 +9,41 @@ import numpy as np from pathlib import Path import math + +SCENARIO_VARIANT_SUFFIXES = ["-H_A2A", "-A2A", "-H_A2A", "-MCP"] + + +def parse_scenario_name(scenario_name: str): + """Return (base_project, variant_suffix). + + variant_suffix is one of: "", "-MCP", "-A2A", "-H_A2A". + """ + for suffix in SCENARIO_VARIANT_SUFFIXES: + if scenario_name.endswith(suffix): + return scenario_name[: -len(suffix)], suffix + return scenario_name, "" + + +def discover_projects_for_version_pair( + scenario_data_dict, suffix_a: str, suffix_b: str +): + """Discover base projects that have both variant scenarios present.""" + projects = set() + for scenario_name in scenario_data_dict.keys(): + base, suffix = parse_scenario_name(scenario_name) + if suffix not in ("", "-MCP", "-A2A", "-H_A2A", "-H_A2A"): + continue + projects.add(base) + return sorted( + [ + base + for base in projects + if f"{base}{suffix_a}" in scenario_data_dict + and f"{base}{suffix_b}" in scenario_data_dict + ] + ) + + try: import matplotlib.pyplot as plt import matplotlib as mpl @@ -30,6 +65,8 @@ METRICS = [ "pass_rate", ] +COMPARISON_METRICS = METRICS + ["average_score"] + # Color palette for different models - professional academic colors (Tableau 10 style) MODEL_COLORS = { "GPT-5": "#1f77b4", # Blue @@ -61,6 +98,9 @@ def df_to_markdown(df): for _, row in df.iterrows(): values = [] for val in row: + if pd.isna(val): + values.append("0.00") + continue if isinstance(val, float): values.append(f"{val:.2f}") else: @@ -70,16 +110,35 @@ def df_to_markdown(df): def metric_display_name(metric, base_project, gt_projects): - if metric == "pass_rate" and gt_projects and base_project in gt_projects: - return "average_score" return metric +def get_metric_value_for_series(metric, base_project, metrics_dict, gt_projects=None): + """Get metric value, calculating average_score from 3 metrics with weights. + + average_score = 0.5 * exact_match + 0.25 * precision + 0.25 * recall + """ + if metric == "average_score": + # Calculate weighted average from 3 components + exact_match = metrics_dict.get("exact_match", 0) + precision = metrics_dict.get("precision", 0) + recall = metrics_dict.get("recall", 0) + + # Check if any value is NaN + if pd.isna(exact_match) or pd.isna(precision) or pd.isna(recall): + return None + + # Weighted average: 0.5 for exact_match, 0.25 each for precision and recall + return 0.5 * exact_match + 0.25 * precision + 0.25 * recall + + val = metrics_dict.get(metric, 0) + if pd.isna(val): + return None + return val + + def series_metric_headers(series_name, gt_projects): - rate_label = ( - "average_score" if gt_projects and series_name in gt_projects else "pass_rate" - ) - return [(rate_label if m == "pass_rate" else m) for m in METRICS] + return [m for m in METRICS] def generate_model_pair_comparison( @@ -89,15 +148,10 @@ def generate_model_pair_comparison( lines = [] lines.append(f"# {comparison_name}\n\n") - # Group scenarios by base project name (remove -A2A, -A2A_mix, -MCP suffixes) + # Group scenarios by base project name (remove -A2A, -H_A2A, -MCP suffixes) project_series = {} for scenario_name in data_dict.keys(): - # Extract base project name - base_name = scenario_name - for suffix in ["-A2A_mix", "-A2A", "-MCP"]: - if scenario_name.endswith(suffix): - base_name = scenario_name[: -len(suffix)] - break + base_name, _ = parse_scenario_name(scenario_name) if base_name not in project_series: project_series[base_name] = [] @@ -108,123 +162,76 @@ def generate_model_pair_comparison( for base_name, scenarios in project_series.items(): series_data[base_name] = {model_a: {}, model_b: {}} - for metric in METRICS: + for metric in COMPARISON_METRICS: vals_a = [] vals_b = [] for scenario_name in scenarios: scenario_data = data_dict[scenario_name] if model_a in scenario_data and model_b in scenario_data: - vals_a.append(scenario_data[model_a].get(metric, 0)) - vals_b.append(scenario_data[model_b].get(metric, 0)) + val_a = get_metric_value_for_series( + metric, + base_name, + scenario_data[model_a], + gt_projects=gt_projects, + ) + val_b = get_metric_value_for_series( + metric, + base_name, + scenario_data[model_b], + gt_projects=gt_projects, + ) + if metric == "pass_rate": + if val_a is None or val_b is None: + continue + if metric == "average_score": + if not val_a or not val_b: + continue + vals_a.append(val_a) + vals_b.append(val_b) if vals_a and vals_b: series_data[base_name][model_a][metric] = np.mean(vals_a) series_data[base_name][model_b][metric] = np.mean(vals_b) # Calculate overall averages - overall_avgs = { - metric: {"a": [], "b": []} for metric in METRICS if metric != "pass_rate" - } - overall_rate_avgs = { - "gt": {"a": [], "b": []}, - "non_gt": {"a": [], "b": []}, - } + overall_avgs = {metric: {"a": [], "b": []} for metric in COMPARISON_METRICS} for base_name in series_data.keys(): - for metric in METRICS: + for metric in COMPARISON_METRICS: if ( metric not in series_data[base_name][model_a] or metric not in series_data[base_name][model_b] ): continue - - if metric == "pass_rate": - bucket = "gt" if gt_projects and base_name in gt_projects else "non_gt" - overall_rate_avgs[bucket]["a"].append( - series_data[base_name][model_a][metric] - ) - overall_rate_avgs[bucket]["b"].append( - series_data[base_name][model_b][metric] - ) - continue - overall_avgs[metric]["a"].append(series_data[base_name][model_a][metric]) overall_avgs[metric]["b"].append(series_data[base_name][model_b][metric]) # Add overall summary section lines.append("## Overall Summary (Averaged Across All Project Series)\n\n") + lines.append( + "_Note: pass_rate is averaged only across project series with available success_rate data; average_score is averaged only across project series with available score data._\n\n" + ) lines.append(f"| Metric | {model_a} | {model_b} | Diff (A-B) |\n") lines.append("| --- | --- | --- | --- |\n") - for metric in METRICS: - if metric == "pass_rate": - if overall_rate_avgs["gt"]["a"]: - avg_a = np.mean(overall_rate_avgs["gt"]["a"]) - avg_b = np.mean(overall_rate_avgs["gt"]["b"]) - diff = avg_a - avg_b - lines.append( - f"| average_score | {avg_a:.2f} | {avg_b:.2f} | {diff:+.2f} |\n" - ) - if overall_rate_avgs["non_gt"]["a"]: - avg_a = np.mean(overall_rate_avgs["non_gt"]["a"]) - avg_b = np.mean(overall_rate_avgs["non_gt"]["b"]) - diff = avg_a - avg_b - lines.append( - f"| pass_rate | {avg_a:.2f} | {avg_b:.2f} | {diff:+.2f} |\n" - ) + for metric in COMPARISON_METRICS: + if not overall_avgs[metric]["a"]: continue - - if overall_avgs[metric]["a"]: - avg_a = np.mean(overall_avgs[metric]["a"]) - avg_b = np.mean(overall_avgs[metric]["b"]) - diff = avg_a - avg_b - lines.append(f"| {metric} | {avg_a:.2f} | {avg_b:.2f} | {diff:+.2f} |\n") + avg_a = np.mean(overall_avgs[metric]["a"]) + avg_b = np.mean(overall_avgs[metric]["b"]) + diff = avg_a - avg_b + lines.append(f"| {metric} | {avg_a:.2f} | {avg_b:.2f} | {diff:+.2f} |\n") lines.append("\n---\n\n") # Add per-series comparisons headers = ["Project Series", f"{model_a}", f"{model_b}", "Diff (A-B)"] - for metric in METRICS: - if metric != "pass_rate": - lines.append(f"## {metric}\n\n") - lines.append("| " + " | ".join(headers) + " |\n") - lines.append("| " + " | ".join(["---"] * len(headers)) + " |\n") - - for base_name in sorted(series_data.keys()): - if ( - metric in series_data[base_name][model_a] - and metric in series_data[base_name][model_b] - ): - val_a = series_data[base_name][model_a][metric] - val_b = series_data[base_name][model_b][metric] - diff = val_a - val_b - lines.append( - f"| {base_name} | {val_a:.2f} | {val_b:.2f} | {diff:+.2f} |\n" - ) - lines.append("\n") - continue - - lines.append("## average_score\n\n") + for metric in COMPARISON_METRICS: + lines.append(f"## {metric}\n\n") lines.append("| " + " | ".join(headers) + " |\n") lines.append("| " + " | ".join(["---"] * len(headers)) + " |\n") for base_name in sorted(series_data.keys()): - if gt_projects and base_name not in gt_projects: - continue - if ( - metric in series_data[base_name][model_a] - and metric in series_data[base_name][model_b] + if metric == "average_score" and ( + not gt_projects or base_name not in gt_projects ): - val_a = series_data[base_name][model_a][metric] - val_b = series_data[base_name][model_b][metric] - diff = val_a - val_b - lines.append( - f"| {base_name} | {val_a:.2f} | {val_b:.2f} | {diff:+.2f} |\n" - ) - lines.append("\n") - - lines.append("## pass_rate\n\n") - lines.append("| " + " | ".join(headers) + " |\n") - lines.append("| " + " | ".join(["---"] * len(headers)) + " |\n") - for base_name in sorted(series_data.keys()): - if gt_projects and base_name in gt_projects: continue if ( metric in series_data[base_name][model_a] @@ -262,11 +269,22 @@ def generate_mcp_comparison(base_project, scenario_data_dict, gt_projects=None): all_models = set(mcp_data.keys()) | set(hardcoded_data.keys()) # Calculate overall averages first - overall_avgs = {metric: {"mcp": [], "hard": []} for metric in METRICS} + overall_avgs = {metric: {"mcp": [], "hard": []} for metric in COMPARISON_METRICS} for model in all_models: - for metric in METRICS: - mcp_val = mcp_data.get(model, {}).get(metric, 0) - hard_val = hardcoded_data.get(model, {}).get(metric, 0) + for metric in COMPARISON_METRICS: + mcp_val = get_metric_value_for_series( + metric, base_project, mcp_data.get(model, {}), gt_projects=gt_projects + ) + hard_val = get_metric_value_for_series( + metric, + base_project, + hardcoded_data.get(model, {}), + gt_projects=gt_projects, + ) + if metric == "pass_rate" and (mcp_val is None or hard_val is None): + continue + if metric == "average_score" and (not mcp_val or not hard_val): + continue overall_avgs[metric]["mcp"].append(mcp_val) overall_avgs[metric]["hard"].append(hard_val) @@ -274,7 +292,7 @@ def generate_mcp_comparison(base_project, scenario_data_dict, gt_projects=None): lines.append("## Overall Summary (Averaged Across All Models)\n\n") lines.append("| Metric | MCP | Hardcoded | Diff (MCP-Hardcoded) |\n") lines.append("| --- | --- | --- | --- |\n") - for metric in METRICS: + for metric in COMPARISON_METRICS: if overall_avgs[metric]["mcp"]: avg_mcp = np.mean(overall_avgs[metric]["mcp"]) avg_hard = np.mean(overall_avgs[metric]["hard"]) @@ -285,7 +303,7 @@ def generate_mcp_comparison(base_project, scenario_data_dict, gt_projects=None): ) lines.append("\n---\n\n") - for metric in METRICS: + for metric in COMPARISON_METRICS: display_metric = metric_display_name(metric, base_project, gt_projects) lines.append(f"## {display_metric}\n\n") headers = ["Model", "MCP", "Hardcoded", "Diff (MCP-Hardcoded)"] @@ -293,8 +311,15 @@ def generate_mcp_comparison(base_project, scenario_data_dict, gt_projects=None): lines.append("| " + " | ".join(["---"] * len(headers)) + " |\n") for model in sorted(all_models): - mcp_val = mcp_data.get(model, {}).get(metric, 0) - hard_val = hardcoded_data.get(model, {}).get(metric, 0) + mcp_val = get_metric_value_for_series( + metric, base_project, mcp_data.get(model, {}), gt_projects=gt_projects + ) + hard_val = get_metric_value_for_series( + metric, + base_project, + hardcoded_data.get(model, {}), + gt_projects=gt_projects, + ) diff = mcp_val - hard_val lines.append( f"| {model} | {mcp_val:.2f} | {hard_val:.2f} | {diff:+.2f} |\n" @@ -308,19 +333,11 @@ def generate_mcp_overall_comparison(projects, scenario_data_dict, gt_projects=No """Generate overall comparison across all MCP vs hardcoded projects.""" lines = [] lines.append("# Overall MCP vs Hardcoded Comparison\n\n") - lines.append( - "Averaged across all projects: MarkdownValidator, GameBuilder, EmailResponder\n\n" - ) + lines.append(f"Averaged across all projects: {', '.join(projects)}\n\n") # Collect data from all projects overall_data = {} - framework_data = { - metric: {"mcp": [], "hard": []} for metric in METRICS if metric != "pass_rate" - } - framework_rate_data = { - "gt": {"mcp": [], "hard": []}, - "non_gt": {"mcp": [], "hard": []}, - } + framework_data = {metric: {"mcp": [], "hard": []} for metric in COMPARISON_METRICS} overall_rate_data = {} for project in projects: @@ -353,52 +370,37 @@ def generate_mcp_overall_comparison(projects, scenario_data_dict, gt_projects=No } for metric in METRICS: - mcp_val = mcp_data.get(model, {}).get(metric, 0) - hard_val = hardcoded_data.get(model, {}).get(metric, 0) - if metric == "pass_rate": - overall_rate_data[model][project_bucket]["mcp"].append(mcp_val) - overall_rate_data[model][project_bucket]["hard"].append(hard_val) - framework_rate_data[project_bucket]["mcp"].append(mcp_val) - framework_rate_data[project_bucket]["hard"].append(hard_val) - else: - overall_data[model][metric]["mcp"].append(mcp_val) - overall_data[model][metric]["hard"].append(hard_val) - framework_data[metric]["mcp"].append(mcp_val) - framework_data[metric]["hard"].append(hard_val) + mcp_val = get_metric_value_for_series( + metric, project, mcp_data.get(model, {}), gt_projects=gt_projects + ) + hard_val = get_metric_value_for_series( + metric, + project, + hardcoded_data.get(model, {}), + gt_projects=gt_projects, + ) + if metric not in overall_data[model]: + overall_data[model][metric] = {"mcp": [], "hard": []} + overall_data[model][metric]["mcp"].append(mcp_val) + overall_data[model][metric]["hard"].append(hard_val) + framework_data[metric]["mcp"].append(mcp_val) + framework_data[metric]["hard"].append(hard_val) # Add framework-level comparison (all models averaged) lines.append("## Framework-Level Comparison (All Models Averaged)\n\n") lines.append("| Metric | MCP | Hardcoded | Diff (MCP-Hardcoded) |\n") lines.append("| --- | --- | --- | --- |\n") - for metric in METRICS: - if metric == "pass_rate": - if framework_rate_data["gt"]["mcp"]: - avg_mcp = np.mean(framework_rate_data["gt"]["mcp"]) - avg_hard = np.mean(framework_rate_data["gt"]["hard"]) - diff = avg_mcp - avg_hard - lines.append( - f"| average_score | {avg_mcp:.2f} | {avg_hard:.2f} | {diff:+.2f} |\n" - ) - if framework_rate_data["non_gt"]["mcp"]: - avg_mcp = np.mean(framework_rate_data["non_gt"]["mcp"]) - avg_hard = np.mean(framework_rate_data["non_gt"]["hard"]) - diff = avg_mcp - avg_hard - lines.append( - f"| pass_rate | {avg_mcp:.2f} | {avg_hard:.2f} | {diff:+.2f} |\n" - ) + for metric in COMPARISON_METRICS: + if not framework_data[metric]["mcp"]: continue - - if framework_data[metric]["mcp"]: - avg_mcp = np.mean(framework_data[metric]["mcp"]) - avg_hard = np.mean(framework_data[metric]["hard"]) - diff = avg_mcp - avg_hard - lines.append( - f"| {metric} | {avg_mcp:.2f} | {avg_hard:.2f} | {diff:+.2f} |\n" - ) + avg_mcp = np.mean(framework_data[metric]["mcp"]) + avg_hard = np.mean(framework_data[metric]["hard"]) + diff = avg_mcp - avg_hard + lines.append(f"| {metric} | {avg_mcp:.2f} | {avg_hard:.2f} | {diff:+.2f} |\n") lines.append("\n---\n\n") # Generate per-model summary tables - for metric in METRICS: + for metric in COMPARISON_METRICS: if metric != "pass_rate": lines.append(f"## {metric}\n\n") headers = [ @@ -411,7 +413,7 @@ def generate_mcp_overall_comparison(projects, scenario_data_dict, gt_projects=No lines.append("| " + " | ".join(["---"] * len(headers)) + " |\n") for model in sorted(overall_data.keys()): - if overall_data[model][metric]["mcp"]: + if overall_data[model].get(metric, {}).get("mcp"): avg_mcp = np.mean(overall_data[model][metric]["mcp"]) avg_hard = np.mean(overall_data[model][metric]["hard"]) diff = avg_mcp - avg_hard @@ -419,34 +421,6 @@ def generate_mcp_overall_comparison(projects, scenario_data_dict, gt_projects=No f"| {model} | {avg_mcp:.2f} | {avg_hard:.2f} | {diff:+.2f} |\n" ) lines.append("\n") - continue - - lines.append("## average_score\n\n") - headers = ["Model", "MCP (Avg)", "Hardcoded (Avg)", "Diff (MCP-Hardcoded)"] - lines.append("| " + " | ".join(headers) + " |\n") - lines.append("| " + " | ".join(["---"] * len(headers)) + " |\n") - for model in sorted(overall_rate_data.keys()): - if overall_rate_data[model]["gt"]["mcp"]: - avg_mcp = np.mean(overall_rate_data[model]["gt"]["mcp"]) - avg_hard = np.mean(overall_rate_data[model]["gt"]["hard"]) - diff = avg_mcp - avg_hard - lines.append( - f"| {model} | {avg_mcp:.2f} | {avg_hard:.2f} | {diff:+.2f} |\n" - ) - lines.append("\n") - - lines.append("## pass_rate\n\n") - lines.append("| " + " | ".join(headers) + " |\n") - lines.append("| " + " | ".join(["---"] * len(headers)) + " |\n") - for model in sorted(overall_rate_data.keys()): - if overall_rate_data[model]["non_gt"]["mcp"]: - avg_mcp = np.mean(overall_rate_data[model]["non_gt"]["mcp"]) - avg_hard = np.mean(overall_rate_data[model]["non_gt"]["hard"]) - diff = avg_mcp - avg_hard - lines.append( - f"| {model} | {avg_mcp:.2f} | {avg_hard:.2f} | {diff:+.2f} |\n" - ) - lines.append("\n") lines.append("---\n\n") return "".join(lines) @@ -516,8 +490,12 @@ def generate_version_comparison( lines.append("| " + " | ".join(["---"] * len(headers)) + " |\n") for model in sorted(all_models): - val_a = data_a.get(model, {}).get(metric, 0) - val_b = data_b.get(model, {}).get(metric, 0) + val_a = get_metric_value_for_series( + metric, base_project, data_a.get(model, {}), gt_projects=gt_projects + ) + val_b = get_metric_value_for_series( + metric, base_project, data_b.get(model, {}), gt_projects=gt_projects + ) diff = val_a - val_b lines.append(f"| {model} | {val_a:.2f} | {val_b:.2f} | {diff:+.2f} |\n") lines.append("\n") @@ -578,8 +556,12 @@ def generate_version_overall_comparison( } for metric in METRICS: - val_a = data_a.get(model, {}).get(metric, 0) - val_b = data_b.get(model, {}).get(metric, 0) + val_a = get_metric_value_for_series( + metric, project, data_a.get(model, {}), gt_projects=gt_projects + ) + val_b = get_metric_value_for_series( + metric, project, data_b.get(model, {}), gt_projects=gt_projects + ) if metric == "pass_rate": overall_rate_data[model][project_bucket]["a"].append(val_a) overall_rate_data[model][project_bucket]["b"].append(val_b) @@ -862,8 +844,10 @@ def create_legend_pdf_horizontal(output_path): def main(): """Main function to process all subfolders and generate radar charts.""" - base_dir = Path("/Users/wzr/TOSEM-2025/RESULTS/RQ1") - output_dir = base_dir / "RadarCharts" + base_dir = Path(__file__).resolve().parent + if not list(base_dir.glob("*/evaluation_results.csv")): + base_dir = Path("/Users/wzr/TOSEM-2025/RESULTS/RQ1") + output_dir = Path(__file__).resolve().parent / "RadarCharts" output_dir.mkdir(exist_ok=True) # Find all evaluation_results.csv files @@ -886,7 +870,7 @@ def main(): # Process each subfolder for csv_file in sorted(csv_files): scenario_name = csv_file.parent.name - series_name = scenario_name.split("-")[0] + series_name, _ = parse_scenario_name(scenario_name) try: # Read the main evaluation results CSV file @@ -930,29 +914,28 @@ def main(): how="left", ) df = df.drop(columns=["Model"]) - df["pass_rate"] = df["pass_rate"].fillna(0.0) + df["pass_rate"] = df["pass_rate"].astype(float) else: print(f" Warning: success_rate.csv not found for {scenario_name}") - df["pass_rate"] = 0.0 + df["pass_rate"] = np.nan - df["pass_rate_agg"] = df["pass_rate"] + # Calculate average_score from 3 metrics with weights + # average_score = 0.5 * exact_match + 0.25 * precision + 0.25 * recall + df["average_score"] = ( + 0.5 * df["exact_match"] + 0.25 * df["precision"] + 0.25 * df["recall"] + ) + # Determine pass_rate_label based on whether score_summary.csv exists pass_rate_label = "pass_rate" score_file = csv_file.parent / "score_summary.csv" if score_file.exists(): - score_df = pd.read_csv(score_file) - score_df["average_score"] = score_df["Mean_Score"] / 100.0 - df = df.merge( - score_df[["Model", "average_score"]], - left_on="model", - right_on="Model", - how="left", - ) - df["pass_rate"] = df["average_score"].fillna(df["pass_rate"]) - df = df.drop(columns=["Model", "average_score"]) pass_rate_label = "average_score" gt_projects.add(series_name) + df["rate_for_plot"] = df["average_score"].where( + df["average_score"] > 0, df["pass_rate"] + ) + # Check if all required metrics are present missing_metrics = [m for m in METRICS if m not in df.columns] if missing_metrics: @@ -966,6 +949,9 @@ def main(): scenario_data[scenario_name][model_name] = { metric: row[metric] for metric in METRICS } + scenario_data[scenario_name][model_name]["average_score"] = row[ + "average_score" + ] # Accumulate data for overall and series-level charts for idx, row in df.iterrows(): @@ -983,8 +969,10 @@ def main(): output_path = output_dir / f"{scenario_name}_radar.pdf" # Generate radar chart + df_plot = df.copy() + df_plot["pass_rate"] = df_plot["rate_for_plot"].fillna(0.0) create_radar_chart( - df, + df_plot, scenario_name, output_path, pass_rate_label=pass_rate_label, @@ -1159,22 +1147,22 @@ def main(): combined_open_closed_lines.append("\n") # Combine generated markdown files into a single markdown with titles - if combined_open_closed_lines: - combined_md_path = output_dir / "All_Open_vs_Closed_summaries.md" - combined_md_path.write_text( - "".join(combined_open_closed_lines), - encoding="utf-8", - ) - print(f" Combined markdown created: {combined_md_path}") + # if combined_open_closed_lines: + # combined_md_path = output_dir / "All_Open_vs_Closed_summaries.md" + # combined_md_path.write_text( + # "".join(combined_open_closed_lines), + # encoding="utf-8", + # ) + # print(f" Combined markdown created: {combined_md_path}") # Combine all Series_Average_radar.csv into a single markdown with titles - if combined_series_avg_lines: - combined_csv_md_path = output_dir / "All_Series_Average_tables.md" - combined_csv_md_path.write_text( - "".join(combined_series_avg_lines), - encoding="utf-8", - ) - print(f" Combined Series Average markdown created: {combined_csv_md_path}") + # if combined_series_avg_lines: + # combined_csv_md_path = output_dir / "All_Series_Average_tables.md" + # combined_csv_md_path.write_text( + # "".join(combined_series_avg_lines), + # encoding="utf-8", + # ) + # print(f" Combined Series Average markdown created: {combined_csv_md_path}") # Generate overall chart overall_output = output_dir / "Overall_Average_radar.pdf" @@ -1199,9 +1187,9 @@ def main(): gt_projects=gt_projects, ) ) - comparison1_md_path = output_dir / "Comparison_Powerful_vs_Lightweight.md" - comparison1_md_path.write_text("".join(comparison1_lines), encoding="utf-8") - print(f" Powerful vs Lightweight comparison created: {comparison1_md_path}") + # comparison1_md_path = output_dir / "Comparison_Powerful_vs_Lightweight.md" + # comparison1_md_path.write_text("".join(comparison1_lines), encoding="utf-8") + # print(f" Powerful vs Lightweight comparison created: {comparison1_md_path}") # File 2: Reasoning vs Non-Reasoning models comparison2_lines = [] @@ -1224,13 +1212,15 @@ def main(): gt_projects=gt_projects, ) ) - comparison2_md_path = output_dir / "Comparison_Reasoning_vs_NonReasoning.md" - comparison2_md_path.write_text("".join(comparison2_lines), encoding="utf-8") - print(f" Reasoning vs Non-Reasoning comparison created: {comparison2_md_path}") + # comparison2_md_path = output_dir / "Comparison_Reasoning_vs_NonReasoning.md" + # comparison2_md_path.write_text("".join(comparison2_lines), encoding="utf-8") + # print( + # f" Reasoning vs Non-Reasoning comparison created: {comparison2_md_path}" + # ) # File 3: MCP vs Hardcoded comparisons comparison3_lines = [] - projects = ["MarkdownValidator", "GameBuilder", "EmailResponder"] + projects = discover_projects_for_version_pair(scenario_data, "-MCP", "") # Add overall summary first comparison3_lines.append( @@ -1245,19 +1235,15 @@ def main(): generate_mcp_comparison(project, scenario_data, gt_projects=gt_projects) ) - comparison3_md_path = output_dir / "Comparison_MCP_vs_Hardcoded.md" - comparison3_md_path.write_text("".join(comparison3_lines), encoding="utf-8") - print(f" MCP vs Hardcoded comparison created: {comparison3_md_path}") + # comparison3_md_path = output_dir / "Comparison_MCP_vs_Hardcoded.md" + # comparison3_md_path.write_text("".join(comparison3_lines), encoding="utf-8") + # print(f" MCP vs Hardcoded comparison created: {comparison3_md_path}") # File 4: MCP vs A2A comparisons comparison4_lines = [] - version_projects = [ - "SQL_assistant", - "intelligent_recruitment_platform", - "landing_page_generator", - "self_evaluation_loop_flow", - "write_a_book_with_flows", - ] + version_projects = discover_projects_for_version_pair( + scenario_data, "-MCP", "-A2A" + ) # Add overall summary first comparison4_lines.append( @@ -1287,23 +1273,27 @@ def main(): ) ) - comparison4_md_path = output_dir / "Comparison_MCP_vs_A2A.md" - comparison4_md_path.write_text("".join(comparison4_lines), encoding="utf-8") - print(f" MCP vs A2A comparison created: {comparison4_md_path}") + # comparison4_md_path = output_dir / "Comparison_MCP_vs_A2A.md" + # comparison4_md_path.write_text("".join(comparison4_lines), encoding="utf-8") + # print(f" MCP vs A2A comparison created: {comparison4_md_path}") - # File 5: A2A vs A2A_mix comparisons + # File 5: A2A vs H_A2A comparisons comparison5_lines = [] + version_projects = discover_projects_for_version_pair( + scenario_data, "-A2A", "-H_A2A" + ) + # Add overall summary first comparison5_lines.append( generate_version_overall_comparison( version_projects, "-A2A", - "-A2A_mix", + "-H_A2A", scenario_data, "A2A", - "A2A_mix", - "A2A vs A2A_mix Comparison", + "H_A2A", + "A2A vs H_A2A Comparison", gt_projects=gt_projects, ) ) @@ -1314,17 +1304,17 @@ def main(): generate_version_comparison( project, "-A2A", - "-A2A_mix", + "-H_A2A", scenario_data, "A2A", - "A2A_mix", + "H_A2A", gt_projects=gt_projects, ) ) - comparison5_md_path = output_dir / "Comparison_A2A_vs_A2A_mix.md" - comparison5_md_path.write_text("".join(comparison5_lines), encoding="utf-8") - print(f" A2A vs A2A_mix comparison created: {comparison5_md_path}") + # comparison5_md_path = output_dir / "Comparison_A2A_vs_H_A2A.md" + # comparison5_md_path.write_text("".join(comparison5_lines), encoding="utf-8") + # print(f" A2A vs H_A2A comparison created: {comparison5_md_path}") print("=" * 60) print(f" Overall average chart created: {overall_output}") diff --git a/data/processed/RQ1/generate_radar_markdowns.py b/data/processed/RQ1/generate_radar_markdowns.py new file mode 100644 index 0000000000000000000000000000000000000000..275db4ffb245eab00829b52616c9982cf2e467f1 --- /dev/null +++ b/data/processed/RQ1/generate_radar_markdowns.py @@ -0,0 +1,1063 @@ +#!/usr/bin/env python3 +"""Generate markdown summaries and comparison tables for RQ1 results. + +This script generates markdown files with proper weighted averaging: +- average_score: Only for GT projects (MarkdownValidator, SQLAssistant, LandingPageGenerator) + Read from score_summary.csv (Mean_Score field) +- pass_rate: For non-GT projects (BookWriter, EmailResponder, GameBuilder, + RecruitmentAssistant, SocialMediaManager) + Read from success_rate.csv (Success_Rate(%) field) +- Other metrics: Read from evaluation_results.csv + +All averages are weighted by sample count (num_samples or Total_Samples). +""" + +import pandas as pd +import numpy as np +from pathlib import Path + +from generate_radar_charts import ( + MODEL_COLORS, + MODEL_CATEGORY, + parse_scenario_name, + discover_projects_for_version_pair, +) + +# GT projects that use score_summary.csv for average_score +GT_PROJECT_PREFIXES = {"MarkdownValidator", "SQLAssistant", "LandingPageGenerator"} + +# Non-GT projects that use success_rate.csv for pass_rate +NON_GT_PROJECT_PREFIXES = { + "BookWriter", + "EmailResponder", + "GameBuilder", + "RecruitmentAssistant", + "SocialMediaManager", +} + +# Metrics that come from evaluation_results.csv (excluding pass_rate) +EVAL_METRICS = [ + "exact_match", + "any_order_match", + "precision", + "recall", + "retry_rate", +] + + +def is_gt_project(series_name: str) -> bool: + """Check if a project series is a GT project (uses average_score).""" + return series_name in GT_PROJECT_PREFIXES + + +def load_scenario_data(csv_file: Path) -> dict: + """Load all data for a single scenario (project folder). + + Returns a dict with: + - 'eval_df': DataFrame from evaluation_results.csv with retry_rate merged + - 'score_df': DataFrame from score_summary.csv (if exists) + - 'success_df': DataFrame from success_rate.csv (if exists) + - 'is_gt': Whether this is a GT project + """ + scenario_name = csv_file.parent.name + series_name, _ = parse_scenario_name(scenario_name) + + result = { + "scenario_name": scenario_name, + "series_name": series_name, + "is_gt": is_gt_project(series_name), + "eval_df": None, + "score_df": None, + "success_df": None, + } + + # Read evaluation_results.csv + df = pd.read_csv(csv_file) + + # Read retry_summary.csv for retry_rate + retry_file = csv_file.parent / "retry_summary.csv" + if retry_file.exists(): + retry_df = pd.read_csv(retry_file) + retry_rate_col = "Retry_Rate(%)" + if ( + retry_rate_col not in retry_df.columns + and "Error_Rate(%)" in retry_df.columns + ): + retry_rate_col = "Error_Rate(%)" + retry_df["retry_rate"] = retry_df[retry_rate_col] / 100.0 + df = df.merge( + retry_df[["Model", "retry_rate"]], + left_on="model", + right_on="Model", + how="left", + ) + df = df.drop(columns=["Model"]) + df["retry_rate"] = df["retry_rate"].fillna(0.0) + else: + df["retry_rate"] = 0.0 + + result["eval_df"] = df + + # Read score_summary.csv for GT projects + score_file = csv_file.parent / "score_summary.csv" + if score_file.exists(): + result["score_df"] = pd.read_csv(score_file) + + # Read success_rate.csv for non-GT projects + success_file = csv_file.parent / "success_rate.csv" + if success_file.exists(): + result["success_df"] = pd.read_csv(success_file) + + return result + + +def weighted_average(values: list, weights: list) -> float: + """Calculate weighted average.""" + if not values or not weights or sum(weights) == 0: + return 0.0 + return sum(v * w for v, w in zip(values, weights)) / sum(weights) + + +def generate_overall_summary_table( + series_averages: dict, + series_weights: dict, +) -> str: + """Generate overall summary table with weighted averages. + + Args: + series_averages: {series_name: {model: {metric: value}}} + series_weights: {series_name: {model: total_samples}} + """ + lines = [] + lines.append("# Overall Summary (All Projects - Weighted Average)\n\n") + lines.append( + "_Note: average_score is only computed for GT projects (MarkdownValidator, SQLAssistant, LandingPageGenerator). pass_rate is only computed for non-GT projects._\n\n" + ) + + # Collect all models + all_models = set() + for series_name, model_data in series_averages.items(): + all_models.update(model_data.keys()) + + # Build data for each model + rows = [] + for model in sorted(all_models): + row = {"model": model} + + # Calculate weighted average for each metric across all series + for metric in EVAL_METRICS: + values = [] + weights = [] + for series_name, model_data in series_averages.items(): + if model in model_data and metric in model_data[model]: + values.append(model_data[model][metric]) + weights.append(series_weights[series_name].get(model, 0)) + row[metric] = weighted_average(values, weights) + + # average_score only from GT projects + avg_score_values = [] + avg_score_weights = [] + for series_name in GT_PROJECT_PREFIXES: + if series_name in series_averages and model in series_averages[series_name]: + if "average_score" in series_averages[series_name][model]: + avg_score_values.append( + series_averages[series_name][model]["average_score"] + ) + avg_score_weights.append(series_weights[series_name].get(model, 0)) + row["average_score"] = weighted_average(avg_score_values, avg_score_weights) + + # pass_rate only from non-GT projects + pass_rate_values = [] + pass_rate_weights = [] + for series_name in NON_GT_PROJECT_PREFIXES: + if series_name in series_averages and model in series_averages[series_name]: + if "pass_rate" in series_averages[series_name][model]: + pass_rate_values.append( + series_averages[series_name][model]["pass_rate"] + ) + pass_rate_weights.append(series_weights[series_name].get(model, 0)) + row["pass_rate"] = weighted_average(pass_rate_values, pass_rate_weights) + + rows.append(row) + + # Create table + headers = ["model"] + EVAL_METRICS + ["average_score", "pass_rate"] + lines.append("| " + " | ".join(headers) + " |\n") + lines.append("| " + " | ".join(["---"] * len(headers)) + " |\n") + + for row in rows: + vals = [row["model"]] + for metric in EVAL_METRICS: + vals.append(f"{row.get(metric, 0):.2f}") + vals.append(f"{row.get('average_score', 0):.2f}") + vals.append(f"{row.get('pass_rate', 0):.2f}") + lines.append("| " + " | ".join(vals) + " |\n") + + return "".join(lines) + + +def generate_series_summary_table( + series_name: str, + model_averages: dict, + is_gt: bool, +) -> str: + """Generate summary table for a single series.""" + lines = [] + lines.append(f"# {series_name} Series Average\n\n") + + if is_gt: + lines.append( + "_This is a GT project series. Using average_score from score_summary.csv._\n\n" + ) + rate_metric = "average_score" + else: + lines.append( + "_This is a non-GT project series. Using pass_rate from success_rate.csv._\n\n" + ) + rate_metric = "pass_rate" + + # Build rows + rows = [] + for model in sorted(model_averages.keys()): + row = {"model": model} + for metric in EVAL_METRICS: + row[metric] = model_averages[model].get(metric, 0) + row[rate_metric] = model_averages[model].get(rate_metric, 0) + rows.append(row) + + # Create table + headers = ["model"] + EVAL_METRICS + [rate_metric] + lines.append("| " + " | ".join(headers) + " |\n") + lines.append("| " + " | ".join(["---"] * len(headers)) + " |\n") + + for row in rows: + vals = [row["model"]] + for metric in EVAL_METRICS: + vals.append(f"{row.get(metric, 0):.2f}") + if rate_metric == "average_score": + vals.append(f"{row.get(rate_metric, 0):.2f}") + else: + vals.append(f"{row.get(rate_metric, 0):.2f}") + lines.append("| " + " | ".join(vals) + " |\n") + + return "".join(lines) + + +def generate_open_vs_closed_table( + title: str, + model_averages: dict, + is_gt: bool = False, + series_name: str = None, +) -> str: + """Generate open vs closed model comparison table.""" + lines = [] + lines.append(f"# {title}\n\n") + + rate_metric = "average_score" if is_gt else "pass_rate" + metrics_to_show = EVAL_METRICS + [rate_metric] + + # Group models by category + group_data = {} + for group_name in ["open", "closed"]: + group_models = [m for m, cat in MODEL_CATEGORY.items() if cat == group_name] + group_values = {metric: [] for metric in metrics_to_show} + model_count = 0 + + for model in group_models: + if model in model_averages: + model_count += 1 + for metric in metrics_to_show: + if metric in model_averages[model]: + group_values[metric].append(model_averages[model][metric]) + + if model_count > 0: + group_data[group_name] = { + "model_count": model_count, + "metrics": { + metric: np.mean(vals) if vals else 0 + for metric, vals in group_values.items() + }, + } + + if not group_data: + return "" + + # Create table + headers = ["Group", "Model Count"] + metrics_to_show + lines.append("| " + " | ".join(headers) + " |\n") + lines.append("| " + " | ".join(["---"] * len(headers)) + " |\n") + + for group_name in ["open", "closed"]: + if group_name not in group_data: + continue + data = group_data[group_name] + vals = [group_name, str(data["model_count"])] + for metric in metrics_to_show: + if metric == "average_score": + vals.append(f"{data['metrics'].get(metric, 0):.2f}") + else: + vals.append(f"{data['metrics'].get(metric, 0):.2f}") + lines.append("| " + " | ".join(vals) + " |\n") + + lines.append("\n") + return "".join(lines) + + +def generate_model_pair_comparison( + model_a: str, + model_b: str, + scenario_data: dict, + comparison_name: str, + gt_projects: set, +) -> str: + """Generate comparison table between two models across all scenarios.""" + lines = [] + lines.append(f"# {comparison_name}\n\n") + + # Group scenarios by base project name + project_series = {} + for scenario_name in scenario_data.keys(): + base_name, _ = parse_scenario_name(scenario_name) + if base_name not in project_series: + project_series[base_name] = [] + project_series[base_name].append(scenario_name) + + # Calculate averaged metrics for each project series (weighted) + series_data = {} + for base_name, scenarios in project_series.items(): + series_data[base_name] = {model_a: {}, model_b: {}} + is_gt = base_name in gt_projects + + rate_metric = "average_score" if is_gt else "pass_rate" + all_metrics = EVAL_METRICS + [rate_metric] + + for metric in all_metrics: + vals_a = [] + vals_b = [] + weights_a = [] + weights_b = [] + + for scenario_name in scenarios: + if scenario_name not in scenario_data: + continue + sdata = scenario_data[scenario_name] + + if model_a in sdata and model_b in sdata: + if metric in sdata[model_a] and metric in sdata[model_b]: + val_a = sdata[model_a].get(metric) + val_b = sdata[model_b].get(metric) + weight_a = sdata[model_a].get("num_samples", 1) + weight_b = sdata[model_b].get("num_samples", 1) + + if val_a is not None and val_b is not None: + vals_a.append(val_a) + vals_b.append(val_b) + weights_a.append(weight_a) + weights_b.append(weight_b) + + if vals_a and vals_b: + series_data[base_name][model_a][metric] = weighted_average( + vals_a, weights_a + ) + series_data[base_name][model_b][metric] = weighted_average( + vals_b, weights_b + ) + + # Calculate overall averages (weighted by series total samples) + overall_avgs = { + metric: {"a": [], "b": [], "wa": [], "wb": []} + for metric in EVAL_METRICS + ["average_score", "pass_rate"] + } + + for base_name in series_data.keys(): + is_gt = base_name in gt_projects + rate_metric = "average_score" if is_gt else "pass_rate" + + for metric in EVAL_METRICS: + if ( + metric in series_data[base_name][model_a] + and metric in series_data[base_name][model_b] + ): + overall_avgs[metric]["a"].append( + series_data[base_name][model_a][metric] + ) + overall_avgs[metric]["b"].append( + series_data[base_name][model_b][metric] + ) + + if ( + rate_metric in series_data[base_name][model_a] + and rate_metric in series_data[base_name][model_b] + ): + overall_avgs[rate_metric]["a"].append( + series_data[base_name][model_a][rate_metric] + ) + overall_avgs[rate_metric]["b"].append( + series_data[base_name][model_b][rate_metric] + ) + + # Add overall summary section + lines.append("## Overall Summary (Averaged Across All Project Series)\n\n") + lines.append( + "_Note: average_score is averaged only across GT projects; pass_rate is averaged only across non-GT projects._\n\n" + ) + lines.append(f"| Metric | {model_a} | {model_b} | Diff (A-B) |\n") + lines.append("| --- | --- | --- | --- |\n") + + for metric in EVAL_METRICS + ["average_score", "pass_rate"]: + if not overall_avgs[metric]["a"]: + continue + avg_a = np.mean(overall_avgs[metric]["a"]) + avg_b = np.mean(overall_avgs[metric]["b"]) + diff = avg_a - avg_b + if metric == "average_score": + lines.append(f"| {metric} | {avg_a:.2f} | {avg_b:.2f} | {diff:+.2f} |\n") + else: + lines.append(f"| {metric} | {avg_a:.2f} | {avg_b:.2f} | {diff:+.2f} |\n") + + lines.append("\n---\n\n") + + # Add per-series comparisons for each metric + for metric in EVAL_METRICS + ["average_score", "pass_rate"]: + lines.append(f"## {metric}\n\n") + headers = ["Project Series", f"{model_a}", f"{model_b}", "Diff (A-B)"] + lines.append("| " + " | ".join(headers) + " |\n") + lines.append("| " + " | ".join(["---"] * len(headers)) + " |\n") + + for base_name in sorted(series_data.keys()): + is_gt = base_name in gt_projects + + # For average_score, only show GT projects + if metric == "average_score" and not is_gt: + continue + # For pass_rate, only show non-GT projects + if metric == "pass_rate" and is_gt: + continue + + if ( + metric in series_data[base_name][model_a] + and metric in series_data[base_name][model_b] + ): + val_a = series_data[base_name][model_a][metric] + val_b = series_data[base_name][model_b][metric] + diff = val_a - val_b + if metric == "average_score": + lines.append( + f"| {base_name} | {val_a:.2f} | {val_b:.2f} | {diff:+.2f} |\n" + ) + else: + lines.append( + f"| {base_name} | {val_a:.2f} | {val_b:.2f} | {diff:+.2f} |\n" + ) + + lines.append("\n") + + return "".join(lines) + + +def generate_version_comparison( + base_project: str, + version_a_suffix: str, + version_b_suffix: str, + scenario_data: dict, + version_a_name: str, + version_b_name: str, + gt_projects: set, +) -> str: + """Generate comparison between two versions of a project.""" + lines = [] + lines.append(f"# {base_project}: {version_a_name} vs {version_b_name}\n\n") + + scenario_a = f"{base_project}{version_a_suffix}" + scenario_b = f"{base_project}{version_b_suffix}" + + if scenario_a not in scenario_data or scenario_b not in scenario_data: + lines.append("_Data not available for comparison_\n\n") + return "".join(lines) + + data_a = scenario_data[scenario_a] + data_b = scenario_data[scenario_b] + + all_models = set(data_a.keys()) | set(data_b.keys()) + is_gt = base_project in gt_projects + rate_metric = "average_score" if is_gt else "pass_rate" + all_metrics = EVAL_METRICS + [rate_metric] + + # Calculate overall averages + overall_avgs = {metric: {"a": [], "b": []} for metric in all_metrics} + for model in all_models: + for metric in all_metrics: + val_a = data_a.get(model, {}).get(metric, 0) + val_b = data_b.get(model, {}).get(metric, 0) + if val_a is not None: + overall_avgs[metric]["a"].append(val_a) + if val_b is not None: + overall_avgs[metric]["b"].append(val_b) + + # Add overall summary section + lines.append("## Overall Summary (Averaged Across All Models)\n\n") + lines.append( + f"| Metric | {version_a_name} | {version_b_name} | Diff ({version_a_name}-{version_b_name}) |\n" + ) + lines.append("| --- | --- | --- | --- |\n") + + for metric in all_metrics: + if overall_avgs[metric]["a"]: + avg_a = np.mean(overall_avgs[metric]["a"]) + avg_b = np.mean(overall_avgs[metric]["b"]) + diff = avg_a - avg_b + if metric == "average_score": + lines.append( + f"| {metric} | {avg_a:.2f} | {avg_b:.2f} | {diff:+.2f} |\n" + ) + else: + lines.append( + f"| {metric} | {avg_a:.2f} | {avg_b:.2f} | {diff:+.2f} |\n" + ) + + lines.append("\n---\n\n") + + # Per-metric tables + for metric in all_metrics: + lines.append(f"## {metric}\n\n") + headers = [ + "Model", + version_a_name, + version_b_name, + f"Diff ({version_a_name}-{version_b_name})", + ] + lines.append("| " + " | ".join(headers) + " |\n") + lines.append("| " + " | ".join(["---"] * len(headers)) + " |\n") + + for model in sorted(all_models): + val_a = data_a.get(model, {}).get(metric, 0) or 0 + val_b = data_b.get(model, {}).get(metric, 0) or 0 + diff = val_a - val_b + if metric == "average_score": + lines.append(f"| {model} | {val_a:.2f} | {val_b:.2f} | {diff:+.2f} |\n") + else: + lines.append(f"| {model} | {val_a:.2f} | {val_b:.2f} | {diff:+.2f} |\n") + + lines.append("\n") + + return "".join(lines) + + +def generate_version_overall_comparison( + projects: list, + version_a_suffix: str, + version_b_suffix: str, + scenario_data: dict, + version_a_name: str, + version_b_name: str, + comparison_title: str, + gt_projects: set, +) -> str: + """Generate overall comparison across all projects for two versions.""" + lines = [] + lines.append(f"# Overall {comparison_title}\n\n") + lines.append(f"Averaged across all projects: {', '.join(projects)}\n\n") + + # Collect data from all projects + overall_data = {} # {model: {metric: {"a": [values], "b": [values]}}} + framework_data = { + metric: {"a": [], "b": []} + for metric in EVAL_METRICS + ["average_score", "pass_rate"] + } + + for project in projects: + scenario_a = f"{project}{version_a_suffix}" + scenario_b = f"{project}{version_b_suffix}" + + if scenario_a not in scenario_data or scenario_b not in scenario_data: + continue + + data_a = scenario_data[scenario_a] + data_b = scenario_data[scenario_b] + + all_models = set(data_a.keys()) | set(data_b.keys()) + is_gt = project in gt_projects + rate_metric = "average_score" if is_gt else "pass_rate" + + for model in all_models: + if model not in overall_data: + overall_data[model] = { + m: {"a": [], "b": []} + for m in EVAL_METRICS + ["average_score", "pass_rate"] + } + + for metric in EVAL_METRICS: + val_a = data_a.get(model, {}).get(metric) + val_b = data_b.get(model, {}).get(metric) + if val_a is not None: + overall_data[model][metric]["a"].append(val_a) + framework_data[metric]["a"].append(val_a) + if val_b is not None: + overall_data[model][metric]["b"].append(val_b) + framework_data[metric]["b"].append(val_b) + + # Rate metric + val_a = data_a.get(model, {}).get(rate_metric) + val_b = data_b.get(model, {}).get(rate_metric) + if val_a is not None: + overall_data[model][rate_metric]["a"].append(val_a) + framework_data[rate_metric]["a"].append(val_a) + if val_b is not None: + overall_data[model][rate_metric]["b"].append(val_b) + framework_data[rate_metric]["b"].append(val_b) + + # Framework-level comparison + lines.append("## Framework-Level Comparison (All Models Averaged)\n\n") + lines.append( + f"| Metric | {version_a_name} | {version_b_name} | Diff ({version_a_name}-{version_b_name}) |\n" + ) + lines.append("| --- | --- | --- | --- |\n") + + for metric in EVAL_METRICS + ["average_score", "pass_rate"]: + if framework_data[metric]["a"]: + avg_a = np.mean(framework_data[metric]["a"]) + avg_b = np.mean(framework_data[metric]["b"]) + diff = avg_a - avg_b + if metric == "average_score": + lines.append( + f"| {metric} | {avg_a:.2f} | {avg_b:.2f} | {diff:+.2f} |\n" + ) + else: + lines.append( + f"| {metric} | {avg_a:.2f} | {avg_b:.2f} | {diff:+.2f} |\n" + ) + + lines.append("\n---\n\n") + + # Per-model summary tables + for metric in EVAL_METRICS + ["average_score", "pass_rate"]: + lines.append(f"## {metric}\n\n") + headers = [ + "Model", + f"{version_a_name} (Avg)", + f"{version_b_name} (Avg)", + f"Diff ({version_a_name}-{version_b_name})", + ] + lines.append("| " + " | ".join(headers) + " |\n") + lines.append("| " + " | ".join(["---"] * len(headers)) + " |\n") + + for model in sorted(overall_data.keys()): + if overall_data[model][metric]["a"]: + avg_a = np.mean(overall_data[model][metric]["a"]) + avg_b = ( + np.mean(overall_data[model][metric]["b"]) + if overall_data[model][metric]["b"] + else 0 + ) + diff = avg_a - avg_b + if metric == "average_score": + lines.append( + f"| {model} | {avg_a:.2f} | {avg_b:.2f} | {diff:+.2f} |\n" + ) + else: + lines.append( + f"| {model} | {avg_a:.2f} | {avg_b:.2f} | {diff:+.2f} |\n" + ) + + lines.append("\n") + + lines.append("---\n\n") + return "".join(lines) + + +def generate_mcp_comparison( + base_project: str, + scenario_data: dict, + gt_projects: set, +) -> str: + """Generate comparison between MCP and hardcoded versions for a project.""" + return generate_version_comparison( + base_project, + "-MCP", + "", + scenario_data, + "MCP", + "Hardcoded", + gt_projects, + ) + + +def generate_mcp_overall_comparison( + projects: list, + scenario_data: dict, + gt_projects: set, +) -> str: + """Generate overall comparison across all MCP vs hardcoded projects.""" + return generate_version_overall_comparison( + projects, + "-MCP", + "", + scenario_data, + "MCP", + "Hardcoded", + "MCP vs Hardcoded Comparison", + gt_projects, + ) + + +def main() -> None: + """Read evaluation CSVs and generate all markdown reports only.""" + + base_dir = Path(__file__).resolve().parent + if not list(base_dir.glob("*/evaluation_results.csv")): + base_dir = Path("/Users/wzr/TOSEM-2025/RESULTS/RQ1") + + output_dir = Path(__file__).resolve().parent / "RadarCharts" + output_dir.mkdir(exist_ok=True) + + # Find all evaluation_results.csv files + csv_files = list(base_dir.glob("*/evaluation_results.csv")) + print(f"Found {len(csv_files)} evaluation_results.csv files") + print("=" * 60) + + # Data structures for aggregation + # scenario_data[scenario_name][model] = {metric: value, num_samples: n} + scenario_data = {} + # series_data[series_name][model] = {metric: [(value, weight), ...]} + series_raw_data = {} + gt_projects = set() + + # Load all scenario data + for csv_file in sorted(csv_files): + scenario_name = csv_file.parent.name + series_name, _ = parse_scenario_name(scenario_name) + + try: + data = load_scenario_data(csv_file) + eval_df = data["eval_df"] + score_df = data["score_df"] + success_df = data["success_df"] + is_gt = data["is_gt"] + + if is_gt: + gt_projects.add(series_name) + + scenario_data[scenario_name] = {} + + for _, row in eval_df.iterrows(): + model_name = row["model"] + num_samples = row.get("num_samples", 1) + + model_data = { + "num_samples": num_samples, + } + + # Read metrics from evaluation_results.csv + for metric in EVAL_METRICS: + if metric in row: + model_data[metric] = row[metric] + + # Read average_score from score_summary.csv for GT projects + if is_gt and score_df is not None: + score_row = score_df[score_df["Model"] == model_name] + if not score_row.empty: + # Mean_Score is in percentage (0-100), convert to decimal (0-1) + model_data["average_score"] = ( + score_row["Mean_Score"].values[0] / 100.0 + ) + model_data["score_samples"] = score_row["Total_Samples"].values[ + 0 + ] + + # Read pass_rate from success_rate.csv for non-GT projects + if not is_gt and success_df is not None: + success_row = success_df[success_df["Model"] == model_name] + if not success_row.empty: + # Success_Rate(%) is in percentage (0-100), convert to decimal (0-1) + model_data["pass_rate"] = ( + success_row["Success_Rate(%)"].values[0] / 100.0 + ) + model_data["pass_rate_samples"] = success_row["Total"].values[0] + + scenario_data[scenario_name][model_name] = model_data + + # Accumulate raw data for series averaging + if series_name not in series_raw_data: + series_raw_data[series_name] = {} + if model_name not in series_raw_data[series_name]: + series_raw_data[series_name][model_name] = { + m: [] for m in EVAL_METRICS + ["average_score", "pass_rate"] + } + series_raw_data[series_name][model_name]["weights"] = [] + series_raw_data[series_name][model_name]["rate_weights"] = [] + + # Store (value, weight) pairs + for metric in EVAL_METRICS: + if metric in model_data: + series_raw_data[series_name][model_name][metric].append( + (model_data[metric], num_samples) + ) + + if "average_score" in model_data: + weight = model_data.get("score_samples", num_samples) + series_raw_data[series_name][model_name]["average_score"].append( + (model_data["average_score"], weight) + ) + + if "pass_rate" in model_data: + weight = model_data.get("pass_rate_samples", num_samples) + series_raw_data[series_name][model_name]["pass_rate"].append( + (model_data["pass_rate"], weight) + ) + + print(f" Loaded {scenario_name} (GT: {is_gt})") + + except Exception as e: + print(f" Error processing {scenario_name}: {e}") + import traceback + + traceback.print_exc() + + print("=" * 60) + print(f"GT projects: {sorted(gt_projects)}") + print("=" * 60) + + # Calculate weighted series averages + # series_averages[series_name][model] = {metric: weighted_avg} + series_averages = {} + series_weights = {} # Total weights per series per model + + for series_name, models_data in series_raw_data.items(): + series_averages[series_name] = {} + series_weights[series_name] = {} + is_gt = series_name in gt_projects + + for model_name, metric_data in models_data.items(): + series_averages[series_name][model_name] = {} + total_weight = 0 + + for metric in EVAL_METRICS: + if metric_data[metric]: + values = [v for v, w in metric_data[metric]] + weights = [w for v, w in metric_data[metric]] + series_averages[series_name][model_name][metric] = weighted_average( + values, weights + ) + total_weight = max(total_weight, sum(weights)) + + if is_gt and metric_data["average_score"]: + values = [v for v, w in metric_data["average_score"]] + weights = [w for v, w in metric_data["average_score"]] + series_averages[series_name][model_name]["average_score"] = ( + weighted_average(values, weights) + ) + + if not is_gt and metric_data["pass_rate"]: + values = [v for v, w in metric_data["pass_rate"]] + weights = [w for v, w in metric_data["pass_rate"]] + series_averages[series_name][model_name]["pass_rate"] = ( + weighted_average(values, weights) + ) + + series_weights[series_name][model_name] = total_weight + + # Generate markdown files + try: + combined_series_avg_lines = [] + combined_open_closed_lines = [] + + # 1) Overall summary + combined_series_avg_lines.append( + generate_overall_summary_table(series_averages, series_weights) + ) + combined_series_avg_lines.append("\n") + + # 2) Per-series summaries + for series_name in sorted(series_averages.keys()): + is_gt = series_name in gt_projects + combined_series_avg_lines.append( + generate_series_summary_table( + series_name, + series_averages[series_name], + is_gt, + ) + ) + combined_series_avg_lines.append("\n") + + # 3) Overall open vs closed + all_models_avg = {} + for series_name, models_data in series_averages.items(): + for model_name, metrics in models_data.items(): + if model_name not in all_models_avg: + all_models_avg[model_name] = { + m: [] for m in EVAL_METRICS + ["average_score", "pass_rate"] + } + for metric, value in metrics.items(): + all_models_avg[model_name][metric].append(value) + + # Average across all series + overall_model_avgs = {} + for model_name, metric_lists in all_models_avg.items(): + overall_model_avgs[model_name] = {} + for metric, values in metric_lists.items(): + if values: + overall_model_avgs[model_name][metric] = np.mean(values) + + combined_open_closed_lines.append( + generate_open_vs_closed_table( + "Overall Open vs Closed Models", + overall_model_avgs, + is_gt=False, # Use pass_rate column for the overall + ) + ) + + # 4) Per-series open vs closed + for series_name in sorted(series_averages.keys()): + is_gt = series_name in gt_projects + combined_open_closed_lines.append( + generate_open_vs_closed_table( + f"{series_name} Open vs Closed Models", + series_averages[series_name], + is_gt=is_gt, + series_name=series_name, + ) + ) + + # Write combined markdowns + combined_csv_md_path = output_dir / "All_Series_Average_tables.md" + combined_csv_md_path.write_text( + "".join(combined_series_avg_lines), encoding="utf-8" + ) + print(f" Combined Series Average markdown created: {combined_csv_md_path}") + + combined_md_path = output_dir / "All_Open_vs_Closed_summaries.md" + combined_md_path.write_text( + "".join(combined_open_closed_lines), encoding="utf-8" + ) + print(f" Combined Open vs Closed markdown created: {combined_md_path}") + + # 5) Pairwise model comparisons + # File 1: GPT-5 vs GPT-4o-mini + comparison1_lines = generate_model_pair_comparison( + "GPT-5", + "GPT-4o-mini", + scenario_data, + "GPT-5 vs GPT-4o-mini (Powerful vs Lightweight)", + gt_projects, + ) + comparison1_md_path = output_dir / "Comparison_Powerful_vs_Lightweight.md" + comparison1_md_path.write_text(comparison1_lines, encoding="utf-8") + print(f" Powerful vs Lightweight comparison created: {comparison1_md_path}") + + # File 2: Reasoning vs Non-Reasoning models + comparison2_lines = [] + comparison2_lines.append( + generate_model_pair_comparison( + "DeepSeek-R1", + "DeepSeek-V3-1", + scenario_data, + "DeepSeek-R1 vs DeepSeek-V3-1 (Reasoning vs Non-Reasoning)", + gt_projects, + ) + ) + comparison2_lines.append("\n---\n\n") + comparison2_lines.append( + generate_model_pair_comparison( + "Gemini-2.5-flash", + "Gemini-2.5-flash-nothinking", + scenario_data, + "Gemini-2.5-flash vs Gemini-2.5-flash-nothinking (Reasoning vs Non-Reasoning)", + gt_projects, + ) + ) + comparison2_md_path = output_dir / "Comparison_Reasoning_vs_NonReasoning.md" + comparison2_md_path.write_text("".join(comparison2_lines), encoding="utf-8") + print(f" Reasoning vs Non-Reasoning comparison created: {comparison2_md_path}") + + # File 3: MCP vs Hardcoded + comparison3_lines = [] + projects = discover_projects_for_version_pair(scenario_data, "-MCP", "") + comparison3_lines.append( + generate_mcp_overall_comparison(projects, scenario_data, gt_projects) + ) + for project in projects: + comparison3_lines.append( + generate_mcp_comparison(project, scenario_data, gt_projects) + ) + comparison3_md_path = output_dir / "Comparison_MCP_vs_Hardcoded.md" + comparison3_md_path.write_text("".join(comparison3_lines), encoding="utf-8") + print(f" MCP vs Hardcoded comparison created: {comparison3_md_path}") + + # File 4: MCP vs A2A + comparison4_lines = [] + version_projects = discover_projects_for_version_pair( + scenario_data, "-MCP", "-A2A" + ) + comparison4_lines.append( + generate_version_overall_comparison( + version_projects, + "-MCP", + "-A2A", + scenario_data, + "MCP", + "A2A", + "MCP vs A2A Comparison", + gt_projects, + ) + ) + for project in version_projects: + comparison4_lines.append( + generate_version_comparison( + project, + "-MCP", + "-A2A", + scenario_data, + "MCP", + "A2A", + gt_projects, + ) + ) + comparison4_md_path = output_dir / "Comparison_MCP_vs_A2A.md" + comparison4_md_path.write_text("".join(comparison4_lines), encoding="utf-8") + print(f" MCP vs A2A comparison created: {comparison4_md_path}") + + # File 5: A2A vs H_A2A + comparison5_lines = [] + version_projects = discover_projects_for_version_pair( + scenario_data, "-A2A", "-H_A2A" + ) + comparison5_lines.append( + generate_version_overall_comparison( + version_projects, + "-A2A", + "-H_A2A", + scenario_data, + "A2A", + "H_A2A", + "A2A vs H_A2A Comparison", + gt_projects, + ) + ) + for project in version_projects: + comparison5_lines.append( + generate_version_comparison( + project, + "-A2A", + "-H_A2A", + scenario_data, + "A2A", + "H_A2A", + gt_projects, + ) + ) + comparison5_md_path = output_dir / "Comparison_A2A_vs_H_A2A.md" + comparison5_md_path.write_text("".join(comparison5_lines), encoding="utf-8") + print(f" A2A vs H_A2A comparison created: {comparison5_md_path}") + + print("=" * 60) + print("All markdown summaries and comparisons generated!") + + except Exception as e: + print(f" Error generating markdown reports: {e}") + import traceback + + traceback.print_exc() + + +if __name__ == "__main__": + main()