Bidirectional intention slot value cross-correlation task type dialogue understanding system and method

Abstract

The invention discloses a bidirectional intention slot value cross-correlation task type dialogue understanding system and method, and the system comprises a text input module, a word vector module, adialogue semantic understanding encoder module, a dialogue semantic understanding decoder module, and an SF- ID network module, a Softmax classification module and a dialogue comprehension information module, the dialogue semantic comprehension encoder module is used for semantic analysis, and the dialogue semantic comprehension decoder module is used for extracting feature vectors of intentionsand slot values; the SF-ID network module is used for performing deep feature extraction on the feature vector through a mechanism of interaction between intention and a slot value; the Softmax classification module is used for acquiring a preliminary result of intention recognition and slot value filling; and the dialogue understanding information module is used for outputting an output result ofintention identification and slot value extraction. According to the system, the performance of natural semantic comprehension can be greatly improved, and the system is simple and easy to implement.

Description

technical field [0001] The present invention relates to the technical fields of information technology and data business, in particular to a task-based dialogue comprehension system and method for bidirectional cross-correlation of intent slot values. Background technique [0002] In recent years, with the rapid development of artificial intelligence, natural language understanding (NLU) technology, as the basic technical support for many applications, has attracted extensive attention from academia and industry, especially in task-based multi-turn dialogue systems, natural language Understanding, as the core technology of semantic parsing, is the upstream module of dialogue management (dm) and natural language generation (nlg). Its effect and performance directly affect the performance and performance of the entire dialogue system, and its importance is self-evident. As a result, natural language understanding (NLU) has received increasing attention from both industry and a...

AI Technical Summary

Problems solved by technology

The construction of this system is huge and complex, involving word segmentation, part-of-speech tagging, nomenclature recognition, syntactic analysis, etc., and this method has gone through three steps in the middle to get the final result of semantic analysis, and the efficiency is relatively low , does not involve the identification of intentions, nor can it be applied to the actual business scenarios of task-based multi-round dialogue systems

Images