Voice assistant intent understanding method, system, electronic device, and computer-readable medium

Abstract

This invention provides a voice assistant intent understanding method, belonging to the field of automotive technology. It identifies natural language text; generates structured semantics based on the natural language text; fuses the structured semantics to form a semantic structure that conforms to the user's intent; and parses the semantic structure that conforms to the user's intent. This invention utilizes the language understanding capabilities inherent in LLM (Low-Level Learning Model), reducing the workload of sample data creation, model training, and development verification, thus shortening the development cycle. While reducing costs and increasing efficiency, it significantly enhances the NLU's ability to understand multiple intents and multi-turn intents, greatly improving the intelligence and user experience of the voice assistant.

Images