Text relation graph-based multi-text abstract generation method

Abstract

The invention discloses a text relation graph-based multi-text abstract generation method, which comprises three stages: in the first stage, a relation graph among texts is constructed according to all input texts, and feature extraction is performed on the texts; in the second stage, the text relation graph data and the text features generated in the first stage are utilized, and a graph neural network is combined to perform high-order feature extraction. and in the third stage, the text features coded in the first two stages are decoded to generate an abstract. In the second stage, the constructed document relation graph and the document representation encoded by the encoder serve as input of the graph convolutional neural network, forward propagation is carried out, and higher-order text features are extracted, so that each document node in the graph can obtain field node information, and the document representation is enriched. When a plurality of documents are input, the mutual relations among the documents can be effectively captured, and the defect that the relations among texts cannot be fully utilized in a traditional method is overcome.

Description

technical field [0001] The invention belongs to the field of natural language processing, and more specifically relates to a method and system for generating a multi-text abstract based on a text relational graph. Background technique [0002] With the development of information technology and smart devices, more and more text data are generated in cyberspace, and the problem of text information overload is increasing. At present, we can easily and quickly obtain a large amount of information. The frequency of obtaining information has increased, and at the same time, the difficulty of obtaining key information has also increased. Therefore, it is more and more important to generalize all kinds of texts to easily obtain key information from them. Text summarization is a technique for concisely and accurately summarizing a large amount of text, using a computer to automatically generate a summary of the input text, so that people can easily obtain key information from a larg...

AI Technical Summary

Problems solved by technology

[0006] In view of the above defects or improvement needs of the prior art, the present invention provides a multi-text summarization method based on a text relational graph, the purpose of which is to solve the problem that the network cannot be executed in parallel during training in the existing method based on the RNN class model. , leading to technical problems of inefficiency in practical applications, as well as the technical problems of poor quality summaries generated by the graph attention-based method because the model cannot understand the semantic association of the input text set well, and the combination of single-text summarization The method of the model is very difficult to locate the key information from the input text sequence, and the quality of the generated abstract is low.

Images