Natural language relatedness tool using mined semantic analysis

This invention describes a search tool that figures out how related natural language terms are. It works by first finding concepts directly linked to a term in a large collection of text. Then, it digs deeper using previously discovered rules to find hidden, or 'latent,' connections between concepts. All these direct and hidden connections are combined into a 'concept space' for the term, and the tool uses these spaces to precisely measure how related different terms are.

Why it matters: Filed before the widespread adoption of large language models (LLMs). The complex 'mined semantic analysis' and 'concept association rules' described in 2016 for building concept spaces can now be significantly enhanced or even replaced by the inherent semantic understanding and generative capabilities of modern LLMs.

Status: Not in forceListed as no longer active. The specific reason is not in the record we hold.
How hard to build: SpecializedNLP algorithms, data mining, concept graph construction

3 AI reinvention directions for this patent

See all three

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

Where this one came from

Every morning we go through millions of abandoned patents and pick out 15 worth your time. This was one of today’s.

This one

+10