Intention recognition response method and device, equipment and storage medium

Abstract

The invention discloses an intention recognition response method. The method comprises the following steps: analyzing each round of acquired dialogue into a dialogue sequence containing a single problem sequence and a response sequence, and carrying out multi-round coding to generate a context hidden state and a query vector; analyzing each piece of knowledge information in the constructed external knowledge base into a knowledge triple; performing multi-hop query in the external knowledge base and updating a query vector to generate a global pointer; calculating a query vector, the similarity between the screened knowledge triad and model deterministic knowledge, capturing target knowledge and generating intention perception representation; performing fine-grained decoding and coarse-grained decoding according to the intention perception representation to output a response sequence with the maximum probability; the invention further discloses an intention recognition response device, corresponding equipment and a storage medium. Through the intention reasoning network, the intention perception representation of the concept mark is obtained by capturing the target knowledge, and the user intention can be effectively recognized to generate accurate response.

Description

technical field [0001] The present application relates to the technical field of end-to-end intent identification and response, and more particularly, to an intent identification and response method, apparatus, device, and storage medium. Background technique [0002] Since the early days of artificial intelligence research, people have been committed to developing highly intelligent human-machine dialogue systems. The first-generation dialogue system is mainly a rule-based dialogue system. Its advantages are that the internal logic is transparent and easy to analyze and debug, but it is highly dependent on the manual intervention of experts, and its flexibility and scalability are poor. With the rise of big data technology, a second-generation data-driven dialogue system based on statistical methods that incorporates reinforcement learning has emerged. It is a modular system that avoids high dependence on experts, but the disadvantage is that the model is difficult to maint...

AI Technical Summary

Problems solved by technology

[0007] ②The pipeline of the dialogue system is a pipeline of four modules in series, so the errors of the previous modules will be amplified later, and there is no filtering or noise reduction measures

Since representations obtained in this way are fragmented by context, it is difficult to model specific conceptual notations of deterministic knowledge

On the other hand, the soft-attention mechanism itself is not suitable for performing fine-grained (token-level) multi-hop reasoning, which makes it difficult to capture the real intention of users to generate accurate responses, especially in complex cross-task scenarios where concept shift may occur

Existing attention-based models are often unable to perform such token-level multi-hop reasoning, which hinders them from obtaining accurate responses

Images