This invention describes a computer system that helps people search databases using everyday language. It works by taking what someone types, breaking it down into pieces that fit database rules, and then building a map (a "query graph") based on those pieces and a set of grammar rules. The system then finds the best path through this map to create a proper database query, which it uses to find search results, and can assign importance scores to connections in its map to help choose the best query.
Why it matters: Filed before LLMs were widely capable of direct query generation. Today, LLMs can often generate complex database queries directly from natural language, potentially simplifying or even superseding the need for a manually defined query grammar and finite state machine.
AI gives you a few directions you could take this. Pick one, and we check whether your version is different enough to patent, then write the filing.
Reinvent this with AI