Text abstract and sentiment classification combined training method

Abstract

The invention provides a text abstract and sentiment classification combined training method, which is realized by adopting a text abstract and sentiment classification combined model, and specifically comprises the following steps of: preprocessing a text, and constructing a training set vocabulary; constructing a text abstract model, and carrying out text abstract task pre-training; adding an emotion classification layer on the basis of a text abstract model, constructing a layered end-to-end model, and jointly training emotion classification and text abstract tasks. The invention provides atext abstract and sentiment classification combined training method. Through combined training of the two types of tasks, the content consistency between the generated abstract and the input text canbe improved, the generated abstract can better contain emotion information of the input text, key information of the input text is extracted through the abstract task, and emotion prediction is moreaccurate.

Description

technical field [0001] The invention relates to a text summarization and sentiment classification method in the field of natural language processing, in particular to a joint training method based on text summarization and sentiment classification. Background technique [0002] With the explosive growth of text information in recent years, people are exposed to massive amounts of text information every day, such as news, microblogs, blogs, reports, papers, etc. Text summarization has a wide range of application scenarios. Intuitively, it can be used to generate news titles, paper keywords, abstracts, etc.; broadly speaking, text summarization technology can also be applied to the result optimization of search engines such as Google and Baidu. The task of extracting key information and forming a refined expression can be solved by automatic text summarization technology. The mainstream methods of text summarization are divided into two categories: extractive (Extractive) and...

AI Technical Summary

Problems solved by technology

In the previous research work, the text summarization and sentiment classification tasks were trained separately through the model, so that the joint expression of the two tasks could not be well learned between the two models.

Images