This invention describes a software system that figures out what a user wants from their written statements. It first cleans the statement, then analyzes each word by its grammatical role, how it connects to other words, and its numerical representation of meaning. A neural network uses these details to categorize each word, and from those categories, the system identifies the user's overall intent.
Why it matters: Filed before the widespread adoption of large language models. Modern LLMs can now perform intent identification with greater accuracy and less explicit feature engineering, potentially simplifying the neural network model described.
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