Upload 8 files

LICENSE.txt

@@ -0,0 +1,14 @@
+Creative Commons Attribution 4.0 International (CC BY 4.0)
+
+You are free to:
+• Share — copy and redistribute the material in any medium or format.
+• Adapt — remix, transform, and build upon the material for any purpose, even commercially.
+
+Under the following terms:
+• Attribution — You must give appropriate credit, provide a link to the license,
+  and indicate if changes were made, without suggesting endorsement.
+
+Full license text: https://creativecommons.org/licenses/by/4.0/
+
+© 2025 Tarek Masryo
+This dataset is released under the CC BY 4.0 International license.

README_space.md

@@ -0,0 +1,352 @@
+---
+license: cc-by-4.0
+task_categories:
+- tabular-classification
+- tabular-regression
+- question-answering
+language:
+- en
+tags:
+- rag
+- retrieval-augmented-generation
+- evaluation
+- hallucination
+- meta-modeling
+- risk-scoring
+- logs
+- telemetry
+- tabular-data
+- multi-table
+- machine-learning
+- open-dataset
+- synthetic
+- simulated
+pretty_name: RAG QA Evaluation Logs & Corpus
+size_categories:
+- 100K<n<1M
+---
+
+# 🧠 RAG QA Evaluation Logs & Corpus  
+### Multi-Table RAG Telemetry for Quality, Hallucinations, Latency, and Cost
+
+A multi-table dataset modeling a **production-style RAG (Retrieval-Augmented Generation) system** end-to-end:
+
+- a **document corpus**
+- a **chunk-level index**
+- **retrieval events** per query
+- **QA evaluation runs** with correctness, hallucination, latency, and cost signals
+- **scenario templates** for QA use cases
+- a **data dictionary** documenting every column
+
+All records are **fully synthetic but realistic** system logs and corpus data — designed to look and behave like real RAG telemetry while remaining safe to share and experiment with.
+
+The dataset is intended for **RAG quality analysis**, **meta-modeling**, **hallucination risk scoring**, and **dashboard-style telemetry**.
+
+---
+
+## 🔐 Privacy & Synthetic Data
+
+This dataset is **fully synthetic**.
+
+- No real users, customers, patients, or organisations are represented.  
+- No personally identifiable information (PII) is included.  
+- All IDs, queries, documents, and logs were **programmatically generated** to mimic realistic RAG system behaviour while preserving privacy.
+
+The design aims to balance **realism** (for meaningful analysis and modeling) with **strong privacy guarantees**, making it suitable for open research, teaching, demos, and public dashboards.
+
+---
+
+## 📘 Dataset Overview
+
+| Field | Description |
+|-------|-------------|
+| **Files** | `rag_corpus_documents.csv`, `rag_corpus_chunks.csv`, `rag_qa_eval_runs.csv`, `rag_retrieval_events.csv`, `rag_qa_scenarios.csv`, `rag_qa_data_dictionary.csv` |
+| **Tables** | 6 (documents, chunks, QA runs, retrieval events, data dictionary) |
+| **Total rows (approx.)** | ~103K across all tables (103,273 rows in total) |
+| **Main targets** | `is_correct`, `hallucination_flag`, `faithfulness_label` |
+| **Type** | Multi-table tabular logs + short text fields |
+
+Tables are linked by **stable identifiers** such as `doc_id`, `chunk_id`, `run_id`, `example_id`, and `scenario_id`, making joins explicit and reliable.
+
+---
+
+## 📂 Files
+
+- `rag_corpus_documents.csv` – document-level corpus metadata  
+- `rag_corpus_chunks.csv` – chunk-level index and text content  
+- `rag_qa_eval_runs.csv` – QA runs with labels, metrics, and configurations  
+- `rag_retrieval_events.csv` – per-chunk retrieval telemetry  
+- `rag_qa_scenarios.csv` – scenario-level QA templates and use cases  
+- `rag_qa_data_dictionary.csv` – column-level documentation for all tables  
+
+All files are in **CSV format**, use **snake_case** column names, and are designed to be **ML- and analytics-ready**.
+
+---
+
+## 📊 Table Summary
+
+Approximate size per table:
+
+| Table                    | Rows   | Columns | Granularity                                           |
+|--------------------------|--------|---------|-------------------------------------------------------|
+| `rag_corpus_documents`   | 658    | 19      | One row per document in the RAG corpus               |
+| `rag_corpus_chunks`      | 5,237  | 6       | One row per text chunk derived from a document       |
+| `rag_qa_eval_runs`       | 3,824  | 46      | One row per QA evaluation example                    |
+| `rag_retrieval_events`   | 93,375 | 9       | One row per retrieved chunk for a given QA example   |
+| `rag_qa_scenarios`       | 88     | 11      | One row per scenario-level QA template / use case    |
+| `rag_qa_data_dictionary` | 91     | 5       | One row per column definition across all tables      |
+
+---
+
+## 🧱 Table Structure
+
+### 1️⃣ Document Corpus — `rag_corpus_documents.csv`
+
+High-level view of the RAG knowledge base.
+
+**Granularity:** 1 row = 1 document
+
+Key fields:
+
+- `doc_id` — unique document identifier  
+- `domain` — e.g. `support_faq`, `hr_policies`, `product_docs`, `developer_docs`, `policies`, `financial_reports`, `medical_guides`, `research_papers`, `customer_success`, `data_platform_docs`, `mlops_docs`, `marketing_analytics`  
+- `title` — document title  
+- `source_type` — e.g. `pdf_manual`, `spreadsheet`, `wiki_page`  
+- `language` — currently `en`  
+- `n_sections`, `n_tokens`, `n_chunks`, `avg_chunk_tokens` — structural and size indicators  
+- `created_at_utc`, `last_updated_at_utc` — lifecycle timestamps  
+- `is_active`, `contains_tables` — operational flags  
+- `pii_risk_level`, `security_tier` — risk and access level  
+- `owner_team`, `embedding_model`, `search_index`, `top_keywords` — ownership and indexing metadata  
+
+Use this table to understand **what kind of corpus the RAG system is built on** and how corpus properties relate to downstream performance.
+
+---
+
+### 2️⃣ Chunk Corpus — `rag_corpus_chunks.csv`
+
+What the retriever actually “sees”.
+
+**Granularity:** 1 row = 1 chunk
+
+Key fields:
+
+- `chunk_id` — unique chunk identifier  
+- `doc_id` — foreign key to `rag_corpus_documents.doc_id`  
+- `domain` — propagated from the parent document  
+- `chunk_index` — 0-based position of the chunk within the document  
+- `estimated_tokens` — approximate token length  
+- `chunk_text` — the text content used for retrieval and ranking  
+
+Use this table to rebuild **retrieval candidates**, inspect chunking strategies, and study how content structure affects retrieval.
+
+---
+
+### 3️⃣ QA Evaluation Runs — `rag_qa_eval_runs.csv`
+
+End-to-end evaluation records for **question–answer runs**, including labels and metrics.
+
+**Granularity:** 1 row = 1 QA example (one query, one answer, one configuration)
+
+Key fields:
+
+**Context & content**
+
+- `example_id`, `run_id` — unique identifiers  
+- `domain`, `task_type`, `difficulty` — scenario and question type  
+- `scenario_id` — link to `rag_qa_scenarios`  
+- `query` — user-style question text  
+- `gold_answer` — reference answer  
+- `has_answer_in_corpus` — whether the corpus actually contains sufficient evidence  
+
+**Quality & hallucination signals**
+
+- `is_correct` — main binary correctness flag  
+- `correctness_label` — descriptive view of correctness (e.g. correct / partial / incorrect)  
+- `faithfulness_label` — e.g. faithful / unfaithful / unknown  
+- `hallucination_flag` — binary hallucination indicator  
+- `user_feedback_label` — simplified user-style feedback  
+- `supervising_judge_label` — synthetic “expert” judgement  
+- `is_noanswer_probe` — marks deliberately unanswerable queries  
+
+**Retrieval metrics**
+
+- `retrieval_strategy`, `chunking_strategy`  
+- `n_retrieved_chunks`  
+- `top1_score`, `mean_retrieved_score`  
+- `recall_at_5`, `recall_at_10`, `mrr_at_10`  
+- `has_relevant_in_top5`, `has_relevant_in_top10`  
+
+**Latency & resource usage**
+
+- `latency_ms_retrieval`, `latency_ms_generation`, `total_latency_ms`  
+- `used_long_context_window`, `context_window_tokens`  
+
+**Configuration & cost**
+
+- `embedding_model`, `reranker_model`, `generator_model`  
+- `temperature`, `top_p`, `max_new_tokens`, `stop_reason`  
+- `prompt_tokens`, `answer_tokens`, `total_cost_usd`  
+
+**Supervision & usage**
+
+- `doc_ids_used`, `chunk_ids_used`  
+- `eval_mode`, `created_at_utc`  
+
+This table is the main entry point for **meta-modeling**, **risk scoring**, and **latency–cost–quality tradeoff analysis**.
+
+---
+
+### 4️⃣ Retrieval Events — `rag_retrieval_events.csv`
+
+Per-chunk retrieval telemetry for each QA example.
+
+**Granularity:** 1 row = 1 retrieved chunk for an example
+
+Key fields:
+
+- `run_id`, `example_id` — link back to `rag_qa_eval_runs`  
+- `chunk_id` — link to `rag_corpus_chunks`  
+- `rank` — rank position (1 = top)  
+- `retrieval_score` — retriever score  
+- `is_relevant` — relevance label for this chunk  
+- `domain`, `difficulty`, `retrieval_strategy` — redundant context fields for easier analysis  
+
+Use this table to reconstruct **retrieval lists**, compute custom **ranking metrics**, and explore how ranking quality influences final answers.
+
+---
+
+### 5️⃣ Scenarios — `rag_qa_scenarios.csv`
+
+Scenario-level templates and use cases for QA runs.
+
+**Granularity:** 1 row = 1 scenario
+
+Key fields:
+
+- `scenario_id` — links to `rag_qa_eval_runs.scenario_id`  
+- `domain` — scenario domain (support, HR, finance, medical, developer docs, etc.)  
+- `primary_doc_id` — anchor document for the scenario  
+- `query`, `gold_answer` — canonical scenario-level QA pair  
+- `difficulty_level` — e.g. easy / medium / hard  
+- `scenario_type` — e.g. factual QA, policy lookup, multi-hop reasoning  
+- `use_case` — short description of the business or product scenario  
+- `has_answer_in_corpus` — whether the scenario is designed to be answerable from the corpus  
+- `n_eval_examples`, `is_used_in_eval` — how many QA examples were generated per scenario and whether it appears in the evaluation runs  
+
+This table adds a **narrative layer** on top of the logs, making it easier to build dashboards, teaching materials, or explainability reports.
+
+---
+
+### 6️⃣ Data Dictionary — `rag_qa_data_dictionary.csv`
+
+Column-level documentation across all tables.
+
+**Granularity:** 1 row = 1 column definition
+
+Key fields:
+
+- `table_name` — name of the table the column belongs to  
+- `column_name` — column name in snake_case  
+- `dtype` — high-level type (int, float, bool, category, datetime, text)  
+- `description` — human-readable explanation  
+- `allowed_values` — expected values or ranges where applicable  
+
+Use this file as a **single source of truth** when exploring or building models on top of the dataset.
+
+---
+
+## 🎯 Targets & Tasks
+
+Typical learning targets:
+
+- **`is_correct`** — classification: did the system answer correctly?  
+- **`hallucination_flag`** — classification: is the answer hallucinated?  
+- **`faithfulness_label`** — multi-class view of answer faithfulness  
+
+Paired with rich system signals (retrieval metrics, latency, cost, configuration), these enable:
+
+- **Meta-models** that estimate answer quality before showing it to users  
+- **Risk scores** driving block / escalate / rerun decisions  
+- **Policy design** for when to switch retrieval strategy or model configuration  
+
+---
+
+## 🚀 Example Usage
+
+Using plain `pandas` with the raw files:
+
+```python
+import pandas as pd
+
+base_path = "path/to/data"  # or the local dataset path
+
+docs = pd.read_csv(f"{base_path}/rag_corpus_documents.csv")
+chunks = pd.read_csv(f"{base_path}/rag_corpus_chunks.csv")
+qa_runs = pd.read_csv(f"{base_path}/rag_qa_eval_runs.csv")
+retrieval_events = pd.read_csv(f"{base_path}/rag_retrieval_events.csv")
+scenarios = pd.read_csv(f"{base_path}/rag_qa_scenarios.csv")
+dictionary = pd.read_csv(f"{base_path}/rag_qa_data_dictionary.csv")
+```
+
+Example join: attach the top-ranked retrieved chunk to each QA example:
+
+```python
+top = (
+    retrieval_events.query("rank == 1")
+    .merge(chunks[["chunk_id", "chunk_text"]], on="chunk_id", how="left")
+)
+
+qa_with_top_chunk = qa_runs.merge(
+    top[["run_id", "chunk_text"]],
+    on="run_id",
+    how="left",
+    suffixes=("", "_top_chunk"),
+)
+```
+
+You can then train a simple meta-model on `qa_with_top_chunk` to predict `is_correct` or `hallucination_flag` from retrieval and configuration features.
+
+---
+
+## 🔬 Research & Applications
+
+- **RAG meta-modeling**  
+  - Predict correctness or hallucination risk from retrieval and latency metrics  
+  - Build guardrails that decide when to block, escalate, or rerun answers  
+
+- **Retrieval & ranking analysis**  
+  - Compare retrieval strategies across domains and difficulty levels  
+  - Explore how rank, score, and recall relate to final correctness  
+
+- **Latency & cost trade-offs**  
+  - Study how `total_latency_ms`, `context_window_tokens`, and `total_cost_usd` interact with answer quality  
+  - Prototype “fast vs careful” modes for RAG systems  
+
+- **Teaching & dashboards**  
+  - Demonstrate a realistic RAG pipeline without exposing real logs  
+  - Build dashboards that visualise quality, latency, cost, and configuration over time  
+
+---
+
+## 🧭 Ethical Considerations
+
+- All records are **fully synthetic** system logs and corpus content, not collected from real users or organisations.  
+- No personally identifiable information (PII) is included.  
+- The dataset is intended for **research, teaching, benchmarking, and prototyping**,  
+  not for validating real-world systems in high-stakes domains (e.g. clinical, legal, financial decisions).
+
+---
+
+## 📚 Citation
+
+When using this dataset in research, demos, or teaching material, please cite the dataset URL on Hugging Face and:
+
+> “**RAG QA Evaluation Logs & Corpus — Synthetic Multi-Table Benchmark** by Tarek Masryo”
+
+---
+
+## 📜 License
+
+**CC BY 4.0 (Attribution Required)**  
+You are free to use, share, and modify this dataset, provided that appropriate credit is given.

data/rag_qa_data_dictionary.csv

@@ -0,0 +1,92 @@
+table_name,column_name,dtype,description,allowed_values
+rag_corpus_documents,doc_id,string,Unique identifier for each document.,
+rag_corpus_documents,domain,string,"High level domain or category of the document (support, product_docs, medical_guides, etc.).",
+rag_corpus_documents,title,text,Short title of the document.,
+rag_corpus_documents,source_type,category,"Source type of the document (kb_article, runbook, policy_pdf, report, etc.).",
+rag_corpus_documents,language,string,"Language of the document, usually ISO language code (e.g., en).",
+rag_corpus_documents,n_sections,int,Number of logical sections inside the document.,
+rag_corpus_documents,n_tokens,int,Estimated total token count for the full document.,
+rag_corpus_documents,n_chunks,int,Number of chunks the document is split into for retrieval.,
+rag_corpus_documents,avg_chunk_tokens,float,Average token count per chunk for this document.,
+rag_corpus_documents,created_at_utc,datetime,UTC timestamp when the document was first created in the corpus.,
+rag_corpus_documents,last_updated_at_utc,datetime,UTC timestamp when the document was last updated.,
+rag_corpus_documents,is_active,bool,Whether the document is currently active and used by the RAG system.,True / False
+rag_corpus_documents,contains_tables,bool,Whether the document contains tabular data.,True / False
+rag_corpus_documents,pii_risk_level,category,Qualitative PII risk for this document.,low / medium / high / none
+rag_corpus_documents,security_tier,category,Security classification tier for the document.,public / internal / restricted / confidential
+rag_corpus_documents,embedding_model,string,Name of the embedding model used to embed this document.,
+rag_corpus_documents,owner_team,string,Logical team or function that owns the document content.,
+rag_corpus_documents,search_index,string,Search index or collection name where this document is indexed.,
+rag_corpus_documents,top_keywords,text,"Representative keywords extracted for the document, stored as a short text list.",
+rag_corpus_chunks,chunk_id,string,Unique identifier for each text chunk in the corpus.,
+rag_corpus_chunks,doc_id,string,Identifier of the parent document that this chunk belongs to.,
+rag_corpus_chunks,domain,string,"Domain of the parent document, repeated for convenience.",
+rag_corpus_chunks,chunk_index,int,Index of the chunk within its parent document (0 based).,
+rag_corpus_chunks,estimated_tokens,int,Estimated token count for the chunk text.,
+rag_corpus_chunks,chunk_text,text,Raw text content of the chunk used for retrieval.,
+rag_retrieval_events,run_id,string,Identifier of the QA evaluation run this retrieval event belongs to. Links to rag_qa_eval_runs.run_id.,
+rag_retrieval_events,chunk_id,string,Identifier of the retrieved chunk. Links to rag_corpus_chunks.chunk_id.,
+rag_retrieval_events,rank,int,Rank position of the chunk in the retrieved list (1 = top ranked).,
+rag_retrieval_events,retrieval_score,float,Raw retrieval score for the chunk (higher is more similar or relevant).,
+rag_retrieval_events,is_relevant,int,Whether this chunk is labeled as relevant to the query.,"0 / 1 (0 = not relevant, 1 = relevant)"
+rag_retrieval_events,domain,string,Domain of the query for this retrieval event.,
+rag_retrieval_events,difficulty,category,Difficulty label of the underlying QA example.,easy / medium / hard
+rag_retrieval_events,retrieval_strategy,category,Retrieval strategy used in this run.,bm25 / dense / hybrid / reranked / other
+rag_retrieval_events,example_id,string,Identifier of the QA example (scenario) used for this run.,
+rag_qa_eval_runs,example_id,string,Identifier for the QA example that this run is evaluating.,
+rag_qa_eval_runs,run_id,string,Unique identifier for this evaluation run. Joins with rag_retrieval_events.run_id.,
+rag_qa_eval_runs,domain,string,Domain or topic of the QA example.,
+rag_qa_eval_runs,task_type,string,"High level task type for the run (e.g., qa, summarization, classification).",
+rag_qa_eval_runs,difficulty,category,"Observed difficulty label for the QA example (easy, medium, hard), derived from retrieval quality, hallucination, and correctness.",easy / medium / hard
+rag_qa_eval_runs,query,text,Natural language query or question posed to the RAG system.,
+rag_qa_eval_runs,gold_answer,text,Reference answer used as the gold standard for evaluation.,
+rag_qa_eval_runs,answer_tokens,int,Approximate token count of the model answer.,
+rag_qa_eval_runs,is_correct,int,"Binary correctness label for the final answer (1 = sufficiently correct, 0 = not correct). Coarser, binary view of the same signal represented by correctness_label.","[0, 1]"
+rag_qa_eval_runs,correctness_label,category,"Multi-class correctness label for the final answer, for example correct / partial / incorrect. More fine-grained view of overall correctness than is_correct.",correct / partially_correct / incorrect / unknown
+rag_qa_eval_runs,faithfulness_label,category,"Multi-class faithfulness label capturing how well the answer is grounded in retrieved evidence (e.g., faithful / unfaithful / unknown).",faithful / unfaithful / uncertain
+rag_qa_eval_runs,hallucination_flag,bool,"Binary hallucination label (1 = hallucination present, 0 = no hallucination detected). Related to the more fine-grained faithfulness_label.","[0, 1]"
+rag_qa_eval_runs,retrieval_strategy,category,Retrieval strategy used for this run.,bm25 / dense / hybrid / reranked / other
+rag_qa_eval_runs,chunking_strategy,category,Chunking strategy used when building the corpus.,fixed_size / semantic / sliding_window / other
+rag_qa_eval_runs,n_retrieved_chunks,int,"Total number of chunks returned by the retriever for this query. May be larger than the number of rows stored in rag_retrieval_events, which usually logs only the top-k results for analysis (e.g., top 10).",
+rag_qa_eval_runs,top1_score,float,Retrieval score of the highest ranked chunk in this run.,
+rag_qa_eval_runs,mean_retrieved_score,float,Mean retrieval score across all retrieved chunks for this run.,
+rag_qa_eval_runs,recall_at_5,float,Binary recall@5 of relevant chunks for this QA example.,
+rag_qa_eval_runs,recall_at_10,float,Binary recall@10 of relevant chunks for this QA example.,
+rag_qa_eval_runs,mrr_at_10,float,Mean reciprocal rank@10 for this QA example.,
+rag_qa_eval_runs,used_long_context_window,bool,Whether a long context window model/config was used.,True / False
+rag_qa_eval_runs,context_window_tokens,int,Maximum context window size in tokens used for this run.,
+rag_qa_eval_runs,latency_ms_retrieval,int,Time taken by retrieval in milliseconds.,
+rag_qa_eval_runs,latency_ms_generation,int,Time taken by answer generation in milliseconds.,
+rag_qa_eval_runs,total_latency_ms,int,Total end to end latency in milliseconds (retrieval + generation + overhead).,
+rag_qa_eval_runs,embedding_model,string,Name of the embedding model powering the retriever.,
+rag_qa_eval_runs,reranker_model,string,"Name of the reranker model, if used.",
+rag_qa_eval_runs,doc_ids_used,text,Pipe separated list of document IDs that contributed context in this run.,
+rag_qa_eval_runs,chunk_ids_used,text,Pipe separated list of chunk IDs that contributed context in this run.,
+rag_qa_eval_runs,supervising_judge_label,category,Label from an external or supervising judge model or human.,
+rag_qa_eval_runs,eval_mode,category,Evaluation mode used for this run.,offline_eval / shadow / canary / live
+rag_qa_eval_runs,user_feedback_label,category,Optional user feedback label for this answer.,positive / negative / mixed / none
+rag_qa_eval_runs,created_at_utc,datetime,UTC timestamp when this run record was created.,
+rag_qa_eval_runs,generator_model,string,Name of the LLM / generator model used to produce the answer.,
+rag_qa_eval_runs,temperature,float,Sampling temperature used for generation.,
+rag_qa_eval_runs,top_p,float,Top-p nucleus sampling parameter used for generation.,
+rag_qa_eval_runs,max_new_tokens,int,Maximum number of new tokens allowed for the generated answer.,
+rag_qa_eval_runs,stop_reason,string,"Reason why the generation stopped (length, stop_sequence, end_of_turn, etc.).",
+rag_qa_eval_runs,prompt_tokens,int,Number of tokens in the prompt / input context.,
+rag_qa_eval_runs,total_cost_usd,float,"Approximate total cost of the run in USD, based on token consumption.",
+rag_qa_scenarios,scenario_id,string,"Unique identifier for the QA scenario (SC001, SC002, ...).",
+rag_qa_scenarios,domain,string,"Domain of the scenario, aligned with corpus document domains.",
+rag_qa_scenarios,primary_doc_id,string,Primary document ID that contains the canonical answer content.,
+rag_qa_scenarios,query,text,User facing question or query for this scenario.,
+rag_qa_scenarios,gold_answer,text,"Gold reference answer for this scenario, grounded in the primary document.",
+rag_qa_scenarios,difficulty_level,category,Scenario difficulty level.,easy / medium / hard
+rag_qa_scenarios,scenario_type,category,"Short label describing the scenario type (policy_lookup, troubleshooting, monitoring, etc.).",
+rag_qa_scenarios,use_case,category,"Intended team or function that would typically ask this question (customer_support, clinical_support, etc.).",
+rag_qa_eval_runs,scenario_id,string,Identifier linking each QA example to a high-level scenario in rag_qa_scenarios.,SC001–SC080 (see rag_qa_scenarios).
+rag_qa_scenarios,has_answer_in_corpus,int,"1 if the scenario is constructed such that the answer exists somewhere in the corpus, 0 for explicit no-answer probes.",0 / 1
+rag_qa_eval_runs,has_answer_in_corpus,int,Flag indicating whether the underlying scenario has an answer in the corpus (1) or is a no-answer probe (0).,0 / 1
+rag_qa_eval_runs,is_noanswer_probe,int,Flag marking queries intentionally designed to have no valid answer in the corpus (no-answer probes). Only a small fraction of examples use this mode.,0 / 1
+rag_qa_eval_runs,has_relevant_in_top5,int,Flag indicating whether at least one relevant chunk was retrieved within the top 5 ranks. Derived from relevance labels in rag_retrieval_events.,0 / 1
+rag_qa_eval_runs,has_relevant_in_top10,int,Flag indicating whether at least one relevant chunk was retrieved within the top 10 ranks. Typically derived from recall_at_10.,0 / 1
+rag_qa_eval_runs,answered_without_retrieval,int,Flag set to 1 when the model produced a correct answer even though recall_at_10 = 0 and the answer exists somewhere in the corpus; 0 otherwise.,0 / 1
+rag_qa_scenarios,n_eval_examples,int,Number of QA evaluation examples in rag_qa_eval_runs that reference this scenario_id.,>= 0
+rag_qa_scenarios,is_used_in_eval,int,"1 if this scenario_id appears at least once in rag_qa_eval_runs, 0 otherwise.",0 / 1

data/rag_qa_scenarios.csv

@@ -0,0 +1,89 @@
+scenario_id,domain,primary_doc_id,query,gold_answer,difficulty_level,scenario_type,use_case,has_answer_in_corpus,n_eval_examples,is_used_in_eval
+SC001,support_faq,DOC0001,How can I reset my password if I no longer have access to my email?,"You can reset your password by verifying your identity using a backup phone number or security questions on the account recovery page. If those options are not available, you must contact support and provide your last login details and billing information for manual verification.",medium,troubleshooting,customer_support,1,49,1
+SC002,support_faq,DOC0001,Where can I see all my open support tickets and their current status?,"All open support tickets are listed under the Support Center in the web portal. Navigate to 'My Tickets' to view each ticket's ID, current status, assigned agent, and last update time.",easy,status_lookup,customer_support,1,21,1
+SC003,product_docs,DOC0004,What is the difference between the Standard and Enterprise pricing plans?,"The Standard plan includes core features with usage limits on API calls and storage, while the Enterprise plan adds dedicated support, higher rate limits, custom SLAs, and advanced access control options.",medium,product_pricing,product_management,1,46,1
+SC004,product_docs,DOC0004,How can an admin enable two-factor authentication for all users in the workspace?,"An admin can enforce two-factor authentication by going to the Security settings in the admin console and enabling the 'Require 2FA for all members' option, then saving the policy. Users will be prompted to configure 2FA on their next login.",medium,configuration_step,security_admin,1,46,1
+SC005,developer_docs,DOC0003,How do I authenticate API requests using service-to-service credentials?,Service-to-service requests are authenticated using a signed JWT. The client service signs a JWT with its private key and sends it in the Authorization header as a bearer token. The platform validates the signature using the registered public key and checks the token's expiry and audience.,medium,api_auth,engineering,1,75,1
+SC006,developer_docs,DOC0003,What should I do if I consistently receive 429 rate limit errors from the write API?,"If you receive 429 errors, you should first reduce the frequency of write requests and implement client-side backoff with jitter. You can also batch non-critical operations and check your current plan's rate limits in the dashboard. If the workload is legitimate and sustained, consider upgrading to a higher tier.",hard,error_handling,engineering,1,65,1
+SC007,financial_reports,DOC0100,Which business segment contributed the most to year-over-year revenue growth in the last fiscal year?,"The enterprise subscriptions segment contributed the most to year-over-year revenue growth, driven primarily by higher renewal rates and expansion within existing customers.",medium,financial_analysis,finance_team,1,163,1
+SC008,financial_reports,DOC0100,"How did operating margin change compared to the previous year, and what were the main drivers?","Operating margin improved compared to the previous year due to revenue growth outpacing operating expenses, especially in sales and marketing efficiencies and reduced infrastructure costs per customer.",hard,financial_explanation,executive_reporting,1,70,1
+SC009,medical_guides,DOC0200,Which symptoms indicate that a patient with chest discomfort should be escalated for urgent evaluation?,"Red-flag symptoms include persistent chest pain, shortness of breath at rest, radiation of pain to the arm or jaw, new confusion, or a sudden drop in blood pressure. These signs require urgent clinical evaluation.",medium,risk_screening,clinical_support,1,101,1
+SC010,medical_guides,DOC0201,What is the recommended approach for adjusting dosage in adult patients with reduced kidney function?,"For adult patients with reduced kidney function, dosage should be adjusted based on the estimated glomerular filtration rate (eGFR). Use the dosing table to select the appropriate reduced dose or extended dosing interval, and monitor renal function regularly.",hard,dosage_adjustment,clinical_support,1,23,1
+SC011,hr_policies,DOC0300,What is the standard probation period for new full-time employees?,"The standard probation period for new full-time employees is three to six months, depending on the role and local regulations. During this period, performance and role fit are reviewed before confirming permanent employment.",easy,policy_lookup,hr_team,1,61,1
+SC012,policies,DOC0305,What is the default data retention period for customer activity logs?,"Customer activity logs are retained for 180 days by default, after which they are automatically deleted or anonymized unless a longer retention period is required for legal or compliance reasons.",medium,compliance_policy,security_compliance,1,86,1
+SC013,support_faq,DOC0001,How can I update the phone number linked to my account?,You can update your phone number from the Account Settings page under Profile. After entering the new number you will need to confirm it with a one-time verification code sent by SMS.,easy,account_update,customer_support,1,21,1
+SC014,support_faq,DOC0001,What should I do if I cannot log in even though my credentials are correct?,First check the service status page to confirm there are no ongoing incidents. If the service is healthy clear your browser cache or try a different device and network. If the issue persists contact support with a screenshot and the approximate time of the failed attempts.,medium,troubleshooting,customer_support,1,48,1
+SC015,support_faq,DOC0002,How do I change the language of notifications and emails I receive?,You can change the language of notifications from the Preferences section in your profile. Select your preferred language and save the changes to apply it to emails and in product alerts.,easy,preference_update,customer_support,1,20,1
+SC016,support_faq,DOC0002,Where can I download my monthly invoices for accounting purposes?,Monthly invoices are available in the Billing section of the web portal. You can download them as PDF or CSV for any past billing period that is still within the retention window.,medium,billing_lookup,customer_support,1,47,1
+SC017,support_faq,DOC0002,How can I reopen a ticket that was closed but the problem came back?,If a problem returns you can either create a new ticket and reference the previous ticket ID or reply to the last email of the closed ticket if reopening is allowed by policy. The support team will link the cases and continue the investigation.,medium,ticket_handling,customer_support,1,47,1
+SC018,support_faq,DOC0002,What are the typical response times for standard versus priority support?,Standard support targets an initial response within one business day while priority support targets a response within a few business hours depending on the severity level.,easy,sla_explanation,customer_support,1,19,1
+SC019,product_docs,DOC0004,Which features are only available in the Enterprise plan?,Enterprise customers gain access to advanced role based access control dedicated customer success management longer data retention and extended audit logging.,medium,plan_comparison,product_management,1,46,1
+SC020,product_docs,DOC0004,Can I downgrade from Enterprise to Standard without losing historical data?,You can downgrade to the Standard plan but some features such as extended retention and advanced logs will no longer be available. Historical data older than the Standard retention window may be deleted according to the data retention policy.,hard,plan_change_impact,product_management,1,28,1
+SC021,product_docs,DOC0005,How can I enable single sign on for my organization?,To enable single sign on you must configure a new SAML or OIDC integration in the admin console then register the corresponding application in your identity provider using the provided callback URL and metadata. Test with a small pilot group before enforcing SSO for all users.,hard,configuration_step,security_admin,1,27,1
+SC022,product_docs,DOC0005,What limits apply to file uploads in the current product version?,The current version supports file uploads up to a documented maximum size per file and a daily global quota per workspace. Certain file types may be blocked for security reasons.,medium,limit_lookup,product_management,1,46,1
+SC023,product_docs,DOC0006,How can I export all workspace data before closing my account?,You can export workspace data by using the bulk export tool in the admin portal which generates downloadable archives of content configuration and activity logs within the retention window.,medium,data_export,product_management,1,45,1
+SC024,product_docs,DOC0006,Does the product support sandbox environments separate from production?,The product supports multiple environments including development staging and production. Each environment can have its own configuration and API keys to keep testing isolated from live data.,easy,environment_capability,product_management,1,75,1
+SC025,developer_docs,DOC0003,How do I paginate through large result sets returned by the list endpoint?,Use the pagination parameters described in the API reference such as page size and next token. After each request pass the token from the previous response until no further token is returned.,medium,api_usage,engineering,1,75,1
+SC026,developer_docs,DOC0003,Which HTTP status codes indicate a retryable error versus a permanent failure?,Retryable errors typically include 429 and 5xx responses. Client errors such as 400 or 401 are usually permanent until the request or credentials are corrected.,medium,error_handling,engineering,1,74,1
+SC027,developer_docs,DOC0007,How can I subscribe to webhooks for real time event notifications?,Create a webhook endpoint in your application then register its URL and the events you want to receive in the developer settings. Verify incoming requests by checking the signature header or shared secret.,medium,integration_setup,engineering,1,74,1
+SC028,developer_docs,DOC0007,What is the recommended way to handle idempotency for write requests?,Use an idempotency key header for write requests and reuse the same key when retrying the same logical operation. The server will return the original result for subsequent requests with the same key.,hard,best_practice,engineering,1,64,1
+SC029,developer_docs,DOC0008,How do I authenticate from a background job that does not have a user session?,Background jobs should use service credentials or an API key with restricted scope rather than user tokens. Store secrets securely and rotate them according to the security guidelines.,medium,api_auth,engineering,1,74,1
+SC030,developer_docs,DOC0008,What are the recommended timeout settings for client side API calls?,The client should set reasonable connection and read timeouts rather than relying on defaults. Timeouts should be tuned according to typical response times and the criticality of the operation.,medium,performance_tuning,engineering,1,74,1
+SC031,developer_docs,DOC0009,How can I test the API in a local development environment without affecting production data?,Use the dedicated sandbox or staging environment with separate API keys. Test data created there does not affect production accounts.,easy,environment_usage,engineering,1,80,1
+SC032,developer_docs,DOC0009,Which SDKs are officially maintained and where can I find examples?,The platform provides official SDKs for several languages and lists them in the developer documentation. Each SDK has example code and quick start guides in its repository.,easy,documentation_lookup,engineering,1,79,1
+SC033,financial_reports,DOC0100,How did total revenue grow compared to the previous fiscal year?,Total revenue increased year over year driven primarily by higher subscription revenue and expansion within existing customers.,easy,financial_analysis,finance_team,1,139,1
+SC034,financial_reports,DOC0100,What were the main cost categories that decreased as a percentage of revenue?,Sales and marketing expenses decreased as a percentage of revenue due to improved efficiency while general and administrative costs remained stable.,medium,cost_breakdown,finance_team,1,163,1
+SC035,financial_reports,DOC0101,How did cash flow from operations change relative to last year?,Cash flow from operations improved compared to last year mainly due to higher profitability and better working capital management.,medium,cashflow_analysis,finance_team,1,163,1
+SC036,financial_reports,DOC0101,What risks related to currency fluctuations are highlighted in the financial report?,The report notes that a significant portion of revenue and costs is denominated in foreign currencies and that adverse currency movements could negatively impact reported results.,hard,risk_disclosure,executive_reporting,1,70,1
+SC037,medical_guides,DOC0200,Which vital sign changes are considered early warning signs of clinical deterioration?,Early warning signs include rising respiratory rate falling oxygen saturation persistent tachycardia or hypotension and new confusion or decreased level of consciousness.,medium,risk_screening,clinical_support,1,101,1
+SC038,medical_guides,DOC0200,When should a patient with chest pain be escalated to emergency assessment?,Patients with chest pain combined with shortness of breath hemodynamic instability or radiation of pain to the arm or jaw should be escalated urgently for emergency assessment.,hard,escalation_criteria,clinical_support,1,23,1
+SC039,medical_guides,DOC0201,How should medication dosage be adjusted in elderly patients with multiple comorbidities?,Dosage in elderly patients should be adjusted based on renal and hepatic function comorbidities and the risk of drug interactions. Start with lower doses and titrate cautiously according to the dosing tables.,hard,dosage_adjustment,clinical_support,1,22,1
+SC040,medical_guides,DOC0201,What non pharmacological measures are recommended before starting sleep medication?,Non pharmacological measures include maintaining regular sleep schedules reducing caffeine and screen exposure in the evening and using relaxation techniques.,easy,lifestyle_guidance,clinical_support,1,49,1
+SC041,medical_guides,DOC0202,Which patients should receive additional monitoring after surgery?,Patients with high risk scores significant comorbidities or major procedures should receive enhanced postoperative monitoring according to the risk stratification tables.,medium,monitoring_plan,clinical_support,1,100,1
+SC042,medical_guides,DOC0202,What guidance is provided for managing mild dehydration in adults?,Mild dehydration in adults is usually managed with oral rehydration solutions and close monitoring of fluid intake and output while addressing any underlying cause.,easy,treatment_guidance,clinical_support,1,49,1
+SC043,hr_policies,DOC0300,What is the company policy on remote work eligibility for full time staff?,Remote work eligibility depends on role type and manager approval. Some roles are fully remote while others require hybrid or on site presence according to the policy.,medium,policy_lookup,hr_team,1,182,1
+SC044,hr_policies,DOC0300,How many days of annual leave are granted to new employees?,New employees receive a standard number of annual leave days prorated based on their start date and local regulations.,easy,benefit_lookup,hr_team,1,61,1
+SC045,hr_policies,DOC0301,What steps should a manager follow when handling a formal performance warning?,A manager should document specific performance issues schedule a formal meeting share expectations and timelines for improvement and record the discussion according to HR templates.,hard,process_guidance,hr_team,1,55,1
+SC046,hr_policies,DOC0301,How is parental leave requested and approved?,Employees must submit a parental leave request through the HR system with expected dates. The request is reviewed by HR and the manager to confirm eligibility and coverage.,medium,leave_process,hr_team,1,182,1
+SC047,hr_policies,DOC0302,What is the process for reporting workplace harassment or discrimination?,Employees can report concerns confidentially through designated HR contacts or an ethics hotline. All reports are investigated according to the company investigation procedures.,hard,escalation_process,hr_team,1,54,1
+SC048,policies,DOC0305,Which types of customer data are classified as highly sensitive?,Highly sensitive data includes payment card information government identifiers health information and authentication secrets.,medium,classification_policy,security_compliance,1,86,1
+SC049,policies,DOC0305,What controls are required when sharing data with third party processors?,Data shared with third party processors must be covered by a data processing agreement and limited to the minimum necessary. Processors must meet documented security and privacy standards.,hard,third_party_policy,security_compliance,1,29,1
+SC050,policies,DOC0306,How long are activity logs retained before deletion or anonymization?,Activity logs are retained for the default retention period described in the policy after which they are either deleted or anonymized unless a longer period is required for legal reasons.,medium,retention_policy,security_compliance,1,85,1
+SC051,policies,DOC0306,Which employees are allowed to access production databases?,Access to production databases is restricted to authorized operations and engineering staff with a documented business need and all access is logged and reviewed.,hard,access_control,security_compliance,1,29,1
+SC052,policies,DOC0307,What is the expected response time for critical security incidents?,Critical security incidents must be acknowledged immediately and investigated according to the incident response plan with predefined timelines for containment and communication.,hard,incident_policy,security_compliance,1,28,1
+SC053,analytics_reports,DOC0400,Which customer segment shows the highest churn risk this quarter?,The analytics report indicates that small customers with low product adoption and limited engagement show the highest churn risk.,medium,segment_insight,analytics_team,1,0,0
+SC054,analytics_reports,DOC0400,How did average session duration change compared to the previous release?,Average session duration increased slightly after the release suggesting improved user engagement with the new features.,easy,trend_analysis,analytics_team,1,0,0
+SC055,analytics_reports,DOC0401,Which feature launches correlated with the largest increase in active users?,The report links the increase in active users to the launch of collaboration features and a simplified onboarding flow.,medium,correlation_insight,analytics_team,1,0,0
+SC056,analytics_reports,DOC0401,What is the current conversion rate from free trial to paid subscription?,The current conversion rate from free trial to paid subscription is summarized in the funnel section of the report and has increased compared to the previous period.,easy,kpi_lookup,analytics_team,1,0,0
+SC057,analytics_reports,DOC0402,Which channels drive the highest lifetime value customers?,Customers acquired through organic search and referrals show the highest lifetime value compared to paid acquisition channels.,medium,channel_performance,analytics_team,1,0,0
+SC058,analytics_reports,DOC0402,What recommendations are made to improve retention based on recent cohort analysis?,The report recommends improving onboarding guidance encouraging early feature adoption and re engaging at risk cohorts with targeted campaigns.,hard,recommendation_summary,analytics_team,1,0,0
+SC059,research_papers,DOC0500,What is the main research question addressed by the paper?,The paper investigates how a proposed method improves performance on benchmark datasets compared to existing approaches.,easy,summary,research,1,45,1
+SC060,research_papers,DOC0500,Which datasets were used for evaluation and why were they chosen?,The authors evaluate on several standard datasets selected for their diversity and widespread use in the field to enable fair comparison.,medium,methodology,research,1,83,1
+SC061,research_papers,DOC0501,What limitations does the paper acknowledge about the proposed approach?,The paper notes that the approach has higher computational cost and may not generalize well beyond the settings tested.,hard,limitations,research,1,51,1
+SC062,research_papers,DOC0501,How do the authors suggest extending this work in future research?,The authors suggest exploring larger scale experiments alternative architectures and real world deployments to validate the findings.,medium,future_work,research,1,82,1
+SC063,security_runbooks,DOC0600,What are the first steps when a suspicious login is detected from an unknown location?,The runbook instructs the team to verify the authenticity of the login force a password reset if needed and review recent activity on the account.,medium,incident_response,operations,1,0,0
+SC064,security_runbooks,DOC0600,How should the team respond to a potential credential leak reported by a third party?,The team should validate the report rotate affected credentials review access logs for misuse and notify stakeholders as described in the incident response plan.,hard,incident_response,operations,1,0,0
+SC065,security_runbooks,DOC0601,What actions are recommended when monitoring detects unusual outbound traffic from a server?,Recommended actions include isolating the server from the network collecting forensic data and scanning for malware or unauthorized processes.,hard,investigation_steps,operations,1,0,0
+SC066,security_runbooks,DOC0601,When can a security incident be closed according to the runbook?,An incident can be closed once containment eradication and recovery steps are completed and a post incident review has been documented.,medium,closure_criteria,operations,1,0,0
+SC067,mlops_docs,DOC0700,How is model drift detected in the production monitoring setup?,Model drift is detected by tracking changes in data distributions performance metrics and stability indicators compared to baseline values.,medium,monitoring,mlops_engineering,1,0,0
+SC068,mlops_docs,DOC0700,What triggers a scheduled model retraining according to the MLOps guidelines?,Retraining is triggered when performance drops below agreed thresholds or when significant data distribution shifts are observed.,medium,retraining_policy,mlops_engineering,1,0,0
+SC069,mlops_docs,DOC0701,How are new model versions rolled out to reduce risk of regressions?,New models are rolled out using staged deployment strategies such as shadow mode or canary releases with close monitoring before full rollout.,hard,deployment_strategy,mlops_engineering,1,0,0
+SC070,mlops_docs,DOC0701,What information must be logged for each prediction in production?,Each prediction log should include model version input features output scores and relevant metadata such as time and request identifier.,medium,logging_requirements,mlops_engineering,1,0,0
+SC071,data_platform_docs,DOC0800,How can I check the freshness of a core analytics table?,You can check table freshness using the data catalog or monitoring dashboard which shows the time of the last successful load for each table.,easy,data_freshness,data_platform_team,1,0,0
+SC072,data_platform_docs,DOC0800,What should I do if a scheduled data pipeline fails overnight?,Follow the pipeline runbook by reviewing error logs rerunning the failed task and notifying downstream consumers if service level objectives are at risk.,medium,pipeline_recovery,data_platform_team,1,0,0
+SC073,data_platform_docs,DOC0801,How are access permissions granted for sensitive analytics tables?,Access to sensitive tables is granted based on roles and must be requested through the access management workflow with approval from data owners.,medium,access_management,data_platform_team,1,0,0
+SC074,data_platform_docs,DOC0801,Which naming conventions should new tables follow in the data warehouse?,New tables should follow documented naming conventions including prefixes by domain consistent use of snake case and clear indicators of table grain.,easy,governance,data_platform_team,1,0,0
+SC075,customer_success,DOC0900,How should customer success managers prepare for a quarterly business review?,They should gather usage metrics recent incidents and upcoming roadmap items and prepare recommendations tailored to the customer objectives.,medium,meeting_preparation,customer_success,1,0,0
+SC076,customer_success,DOC0900,What actions are recommended when early signs of churn are detected?,Recommended actions include scheduling a check in call understanding the root causes and proposing a clear plan to address product or value gaps.,hard,churn_risk,customer_success,1,0,0
+SC077,customer_success,DOC0901,How should new stakeholders be onboarded at an existing customer?,New stakeholders should receive a tailored overview of the solution key dashboards and agreed success metrics along with links to training resources.,medium,onboarding,customer_success,1,0,0
+SC078,marketing_analytics,DOC0950,Which campaigns delivered the highest return on ad spend this month?,The report shows that targeted remarketing and branded search campaigns achieved the highest return on ad spend.,medium,campaign_performance,marketing_team,1,0,0
+SC079,marketing_analytics,DOC0950,How has email open rate changed since the new template was introduced?,Email open rates increased after the new template was introduced indicating better subject lines and design.,easy,kpi_trend,marketing_team,1,0,0
+SC080,marketing_analytics,DOC0951,What recommendation does the report make for reallocating budget across channels?,The report recommends shifting budget toward higher performing organic and referral channels while reducing spend on low converting display campaigns.,hard,budget_recommendation,marketing_team,1,0,0
+SC081,developer_docs,,How can I test the API in a local development environment without affecting production data?,NO_ANSWER_IN_CORPUS,easy,no_answer_probe,no_answer_detection,0,3,1
+SC082,financial_reports,,"How did operating margin change compared to the previous year, and what were the main drivers?",NO_ANSWER_IN_CORPUS,hard,no_answer_probe,no_answer_detection,0,3,1
+SC083,hr_policies,,How many days of annual leave are granted to new employees?,NO_ANSWER_IN_CORPUS,easy,no_answer_probe,no_answer_detection,0,3,1
+SC084,medical_guides,,Which symptoms indicate that a patient with chest discomfort should be escalated for urgent evaluation?,NO_ANSWER_IN_CORPUS,medium,no_answer_probe,no_answer_detection,0,3,1
+SC085,policies,,Which employees are allowed to access production databases?,NO_ANSWER_IN_CORPUS,hard,no_answer_probe,no_answer_detection,0,3,1
+SC086,product_docs,,Can I downgrade from Enterprise to Standard without losing historical data?,NO_ANSWER_IN_CORPUS,hard,no_answer_probe,no_answer_detection,0,3,1
+SC087,research_papers,,How do the authors suggest extending this work in future research?,NO_ANSWER_IN_CORPUS,medium,no_answer_probe,no_answer_detection,0,3,1
+SC088,support_faq,,Where can I download my monthly invoices for accounting purposes?,NO_ANSWER_IN_CORPUS,medium,no_answer_probe,no_answer_detection,0,3,1