This invention describes a computer system that helps manage knowledge by building and updating structured knowledge graphs, called ontologies. When a user provides a text document, the system selects an appropriate ontology based on their account settings. It can then update this ontology by identifying new concepts (represented as n-grams) in the document, converting them into numerical representations (embedding vectors), and adding them as new nodes (vertices) to the ontology if they are not already present.
Why it matters: Filed before the widespread availability of highly capable large language models. The generation of embedding vectors and identification of n-grams for ontology construction, central to the claims, is now significantly more efficient and accurate with current LLM capabilities.
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