Text summary generation method, system, terminal and medium based on sentence fusion

Abstract

The invention provides a method, system, terminal and medium for generating text summaries based on sentence fusion. The input text is segmented and segmented, and the index of the sentence where each word is located is recorded; and the representation of the sentence, obtain the sentence vector, and add the separator and terminator at the end of the sentence vector; use the pointer network to extract important sentences, and divide them into groups according to the separator to form multiple sentence groups; for each sentence group Perform fusion modification to obtain text summaries; use cooperative reinforcement learning to alternately train the extraction network and the fusion network to complete the gradient update of the parameters. A maximum two-player game is used to approximate global optimization through a collaborative reinforcement learning approach. The present invention can enhance the interpretability of the deep learning-based text summarization technology, and more intermediate decisions and results can be displayed to users. And by fusing sentences, the conciseness of the generated text is improved.

Description

technical field [0001] The present invention relates to the technical field of natural language processing, and in particular, to a method, system, terminal and medium for generating text summaries based on sentence fusion. Background technique [0002] Automatic text summarization technology is the process of compressing long articles into concise, coherent summaries while preserving the most important content. When a person is summarizing an article, the abstract usually first selects sentences containing the most basic information, and then simplifies relatively independent single sentences, or fuses multiple sentences to fuse relevant information to achieve the purpose of simplicity and coherence . In popular corpora, about 33%-60% of summary sentences are synthesized from two or more article sentences. These statistics shed light on the human behavior of sentence fusion while summarizing, and the need to study the mechanisms behind sentence fusion. Previous technique...

AI Technical Summary

Problems solved by technology

[0003] At present, there is no description or report of the similar technology of the present invention, and no similar data at home and abroad have been collected yet.

Images