Improve Site Search Relevance

Visitors are telling me that the search feature on my website often misses the mark. I want the results they see to feel immediately useful, driven by smarter keyword matching rather than the current basic lookup. Your task is to tune the existing search logic so that every query returns content that truly matches the words entered and ranks it sensibly from most to least relevant.

The site is built with a standard LAMP stack and uses a custom PHP search controller that queries a MySQL database. I can give you full access to a staging copy, sample queries, and the current relevance-scoring formula so you can benchmark your improvements.

Key deliverables:
• Refactor or replace the keyword-matching algorithm to boost precision and recall
• Add stemming, pluralisation handling, and typo tolerance if needed to lift match quality
• Provide before-and-after relevance tests (precision/recall or similar) on a supplied query set
• Document any database changes and new parameters so I can maintain the code going forward

Acceptance criteria: search accuracy scores at least 25 % higher on the test set without increasing response time beyond 200 ms.

If you have experience with ElasticSearch, Sphinx, or clever SQL full-text tricks, let me know—happy to switch technology stacks if it demonstrably improves relevance.