Turn abandoned patents
into your own filing
into your own filing
This invention describes a system that helps people find specific computer code by comparing a text query to existing code. It works by converting both the query and the code into structured diagrams, or graphs, which represent their underlying meaning. The system then uses advanced computer algorithms, specifically neural networks, to measure how similar these meaning-graphs are, helping to retrieve the most relevant code. The claims narrow this to using a constituency tree for the text query and a program graph for the code, generated via semantic parsing.
Why it matters: Filed before large language models (LLMs) became widely capable of advanced semantic parsing and code understanding. The neural network embeddings and semantic parsing algorithms described would be significantly more powerful and accessible today, making the core task of semantically matching text to code much more feasible and accurate.
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
