Turn abandoned patents
into your own filing
into your own filing
This invention describes a system that automatically learns facts from written text, such as internet articles. It uses artificial intelligence, specifically recurrent neural networks, to read sentences, understand their meaning, and identify relationships between different things mentioned. These learned facts are then stored and can be used to answer questions automatically, without human intervention in creating the fact associations. The system can also perform confidence checks on speech recognition or its own semantic processing.
Why it matters: Filed when recurrent neural networks were state-of-the-art for NLP. Today, Transformer-based models offer significantly better performance for text encoding and relation extraction, making the core task of building a robust fact-extraction and query-answering system much more achievable and powerful.
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
