Spaces:
Sleeping
Sleeping
Commit History
-
Remove evaluation scripts and sample data for MentorMe recommendation system
Doanh Van Vu- Deleted `evaluate_recommendations.py`, `evaluation_report.md`, `sample_mentee_evaluation.json`, `sample_mentor_upsert.json`, and related JSON files for unique careers, domains, and skills to streamline the project structure. - This cleanup aligns with recent refactoring efforts and focuses on maintaining only essential components for the recommendation system.5d75507
-
Enhance logging for scoring and reranking processes
Doanh Van Vu- Added detailed logging in `reranker_service.py` to report the ranking of candidates, including mentor ID, rating, and scores. - Improved logging in `scoring.py` to provide insights into the final ranking of mentors, including their scores and applied weights, enhancing traceability and debugging capabilities.3c1cd35 -
Update recommendation weights in settings for improved model performance
Doanh Van Vu- Adjusted `SEMANTIC_WEIGHT` to 1.0 and `RULE_BASED_WEIGHT` to 0.0 in `settings.py` to prioritize semantic recommendations over rule-based ones, enhancing the recommendation system's effectiveness.7e85f61
-
Enhance evaluation metrics and reporting in recommendation system
Doanh Van Vu- Updated `evaluate_recommendations.py` to include Mean Reciprocal Rank (MRR) as a new evaluation metric alongside existing metrics like Precision@K, Recall@K, Hit Rate@K, and NDCG@K. - Modified the evaluation report generation to incorporate MRR results, ensuring comprehensive performance insights. - Revised `evaluation_report.md` and `sample_mentee_evaluation.json` to reflect updated ground truth data and evaluation results for improved accuracy and relevance.dbdb72a
-
Update sample mentee and mentor data for enhanced evaluation
Doanh Van Vu- Revised `sample_mentee_evaluation.json` to reflect updated goals, skills, and career information for mentees, improving alignment with current industry standards. - Expanded `sample_mentor_upsert.json` with new mentor profiles, including detailed bios, career information, and skills to better represent the mentoring landscape. - Enhanced data structure by adding new fields such as `domain_names`, `skill_names`, and `experiences` for both mentees and mentors, facilitating more comprehensive evaluations and recommendations.ee8ceae -
Update evaluation scripts and data structures for MentorMe recommendation system
Doanh Van Vu- Revised the `generate_research_report` function in `evaluate_recommendations.py` to provide a more concise evaluation report format, focusing on essential information. - Enhanced the report generation by updating metric calculations and restructuring the output for clarity. - Added new fields in `sample_mentee_evaluation.json` and `sample_mentor_upsert.json` to improve data representation, including `career_name` and `domain_ids`. - Introduced new JSON files for unique careers, domains, and skills to standardize data across the evaluation process.da6c096
-
Enhance RecommendationRequest model and mentee query text generation
Doanh Van Vu- Added new optional fields to the RecommendationRequest model: `career_name`, `domain_names`, `mentor_domain_names`, and `skill_names` for improved data representation. - Updated the `build_mentee_query_text` function to utilize the new fields, enhancing the clarity of the generated text by including career names and domain names instead of just IDs.a4869e5 -
Add evaluation scripts and sample data for MentorMe recommendation system
Doanh Van Vu- Introduced `evaluate_recommendations.py` for assessing the performance of the recommendation system, including metrics like Precision@K, Recall@K, Hit Rate@K, and NDCG@K. - Created `evaluation_report.md` to document the evaluation results and insights. - Added `sample_mentee_evaluation.json` and `sample_mentor_upsert.json` for testing and demonstration purposes. - Implemented `upsert_mentors.py` to facilitate the upsert of mentor data into the Pinecone vector database.4369e3b
-
Refactor recommendation request and mentee query text handling
Doanh Van Vu- Updated the RecommendationRequest model to remove unnecessary fields and added `mentor_domain_ids` and `career_id` for improved data structure. - Modified the `build_mentee_query_text` function to reflect changes in the mentee data structure, enhancing clarity by including `career_id`, `domain_ids`, and `mentor_domain_ids` while removing references to `desired_skills`, `current_skills`, and `interests`.71703c8 -
Remove ScheduleDto and related availability fields from models and text builder
Doanh Van Vu- Deleted the ScheduleDto class and its references from the MentorUpsertRequest and RecommendationRequest models, simplifying the data structure. - Updated the text building functions to exclude availability-related information, enhancing clarity in mentor and mentee summaries. - This change aligns with recent refactoring efforts to streamline mentor availability handling.769cb8f -
Refactor mentor availability handling across services
Doanh Van Vu- Removed the `available_slots` and `has_availability` fields from the MentorUpsertRequest and RecommendationRequest models, streamlining the data structure. - Updated the PineconeService and RecommendationService to eliminate references to availability metrics in logging and processing. - Simplified scoring logic in the rerank_mentors function by removing the availability score calculation, enhancing performance and clarity. - Adjusted the mentor text building process to exclude available slots information, improving the output format.69934b0 -
Refactor RecommendationService to remove filtering logic
Doanh Van Vu- Removed the filter construction and fallback mechanism from the recommendation process, simplifying the query to Pinecone by applying no filters. - Updated logging to reflect the changes, ensuring clarity in the absence of filters during mentor queries. - Enhanced the overall readability and maintainability of the code by eliminating unnecessary complexity in the filtering logic.7083cd1 -
Update requirements.txt to include protobuf dependency
Doanh Van Vu- Added protobuf version 3.20.0 to the requirements to ensure compatibility with other libraries and features.2dfb75f -
Implement fallback filtering in RecommendationService for mentor queries
Doanh Van Vu- Added logic to handle cases where no mentors are found with strict filters (rating/availability) by introducing a fallback mechanism that relaxes these constraints. - Enhanced logging to provide detailed information about the filtering process and fallback attempts, improving traceability and debugging capabilities. - Updated the filter construction to include logging for each filter applied based on mentee data.f879018 -
Enhance logging in PineconeService for query results and statistics
Doanh Van Vu- Updated the logging mechanism in the PineconeService to provide detailed information about each mentor result, including metadata when available. - Added statistics logging for score metrics (min, max, avg) after query completion to improve performance monitoring and analysis.ff5d801 -
Add test for recommend mentors endpoint and enhance logging in services
Doanh Van Vu- Introduced a new test script `test_api.py` to validate the recommend mentors API endpoint. - Enhanced logging in `embedding_service.py`, `pinecone_service.py`, `recommendation_service.py`, `reranker_service.py`, and `scoring.py` to include timing information for various operations, improving traceability and performance monitoring. - Improved error handling and logging messages for better clarity and debugging support.6a14fa9 -
Initial commit of the MentorMe AI Recommendation Server, including core application structure, environment configuration, Docker setup, and essential services for mentor recommendations and health checks.
Doanh Van Vu1904012 -
initial commit
Verified signature6dc87ae