Dialogue chapter analysis method

Abstract

The invention discloses a dialogue chapter analysis method, medium and equipment, and the method comprises the steps of: obtaining historical dialogue information, and carrying out the preprocessing of the historical dialogue information, so as to generate an initial training sample; generating a first training sample and a second training sample according to the initial training sample; performing model training according to the first training sample to obtain a structure self-sensing teacher model; performing model training according to the second training sample to obtain a structure self-sensing student model, and pulling close the structure relationship of the structure self-sensing teacher model and the structure relationship of the structure self-sensing student model to obtain a final structure self-sensing student model; obtaining to-be-analyzed dialogue information, and inputting the to-be-analyzed dialogue information into the final structure self-sensing student model to generate a chapter relationship between dialogue rounds corresponding to the to-be-analyzed dialogue information. Dialogue information can be automatically analyzed, and error propagation caused by introduction of an explicit prediction result is avoided; and meanwhile, the chapter analysis effect is improved.

Description

technical field [0001] The present invention relates to the technical field of natural language processing, in particular to a dialogue text analysis method, a computer-readable storage medium and a computer device. Background technique [0002] In related technologies, when parsing dialogue chapters, an encoder is often used to learn the representation of each round of dialogue, and then through splicing or addition methods to predict the discourse relationship between two rounds of dialogue, and there is no structural relationship between the chapters in the model In addition, related technologies introduce explicit historical round prediction results to assist the current round of chapter relationship prediction, which has the problem of error propagation, and the final chapter parsing effect is not good. Contents of the invention [0003] The present invention aims to solve one of the technical problems in the above-mentioned technologies at least to a certain extent. ...

AI Technical Summary

Problems solved by technology

[0002] In related technologies, when parsing dialogue chapters, an encoder is often used to learn the representation of each round of dialogue, and then through splicing or addition methods to predict the discourse relationship between two rounds of dialogue, and there is no structural relationship between the chapters in the model In addition, related technologies introduce explicit historical round prediction results to assist the current round of chapter relationship prediction, which has the problem of error propagation, and the final chapter parsing effect is not good.

Images