End-to-end multitask learning dialogue anaphora resolution method and system

Abstract

The invention provides an end-to-end multitask learning dialogue anaphora resolution method and a system. The system comprises a context information representation module, a zero pronoun attention representation module, a depth detection model and a replacement module. The context information representation module is used for preprocessing a historical dialogue and a current dialogue, extracting a candidate word context representation and a pronoun context representation, and carrying out attention weight calculation on the candidate word context representation and the pronoun context representation; the zero pronoun attention representation module is used for further carrying out attention weight calculation on the candidate word context representation and the pronoun context representation; the deep detection model is used for judging whether an anaphora phenomenon exists in a current session, and the replacement module is used for replacing pronouns and zero pronouns with candidate words. According to the method, an end-to-end multi-task deep learning technology is adopted, the resolution task is completed based on attention mechanism representation, the resolution accuracy is improved, complete recovery of anaphora is ensured, and the intellectualization capability of a dialogue system is improved.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to an end-to-end multi-task learning dialogue reference resolution method and system. Background technique [0002] Reference is a common linguistic phenomenon. Pronouns are often used in articles and conversations to refer to a noun that has appeared in previous texts or previous conversations. Although references make articles or conversations more concise and diverse, this type of reference phenomenon is difficult when processed by machines. Problems with unclear references and unclear intentions occur. Therefore, in order to solve such problems and allow machines to better understand, reference resolution is particularly critical. [0003] With the rapid development of artificial intelligence technology, natural language processing technology has gradually deepened into a more detailed field, and anaphora resolution has become a key issue in natural lang...

AI Technical Summary

Problems solved by technology

[0005] (1) In anaphora detection, a large number of methods only use the sentence information of zero pronouns without combining semantic information, resulting in low accuracy in the detection task and accumulation of errors, which affects the resolution effect

[0006] (2) In the resolution of anaphora, when the current method uses sentence information, the weight of all sentence information is the same, and the key information of the sentence is not highlighted, and the digestion rate is low.

[0007] (3) In reference resolution, it is generally divided into reference detection and reference resolution. Due to the accumulation of errors before and after the two single tasks, the overall resolution rate is limited.

Images