The objective of this project is to support the quality assurance and evaluation of a personalized restaurant recommendation system. The student will focus on testing, validating, and analyzing the accuracy of tag-based metadata and recommendation outputs. Rather than altering any core algorithms, the student will conduct structured audits and evaluations to help identify mismatches, anomalies, or inconsistencies in how restaurants are categorized and recommended. This work is crucial in helping improve data hygiene and tagging consistency, supporting future product reliability without impacting live systems or core ML infrastructure.

Expected Outcomes:

• Audit and evaluate the application of a provided tag taxonomy across a dataset of restaurants
• Conduct simulation-based evaluations of search results using predefined queries and metrics (e.g. tag alignment score, diversity checklists)
• Identify mismatches between intended tag use (e.g. “Romantic” or “Work-Friendly”) and actual recommendations returned
• Recommend refinements to tagging conventions or query interpretation logic

Expected Deliverables:

• A tag audit report flagging inconsistent, missing, or misapplied tags, including category-level summaries
• A spreadsheet (evaluation_results.csv) of test queries and observed system outputs, annotated with scoring insights
• 3–5 documented case studies of user scenarios with commentary on recommendation quality and diversity
• A short presentation or written brief summarizing findings and proposing next-step improvements