Factual information coding and evaluation method for shipping news abstract generation

Abstract

The invention discloses a factual information coding and evaluation method for shipping news abstract generation, and the method comprises the following steps: inputting news text information and carrying out semantic coding to obtain multi-level news text coding characteristics and information; constructing a multi-level entity based on the multi-level news text coding features and information, and constructing an entity relation graph based on the multi-level entity; extracting the factual information of different levels in the entity relation graph; splicing elements of the tetrad in the factual information to obtain a factual information code; modeling the process of judging the importance of the factual information into a specific dichotomy task for scoring, selecting multiple pieces of factual information based on scores, and converting the factual information into a factual information graph; traversing the factual information graph through an attention mechanism network and a bidirectional LSTM network to generate factual information of the text abstract; and fusing the factual information of the text abstract into a news abstract generation model, and generating a shipping news abstract through the model. And the abstract generation effect and accuracy are improved.

Description

technical field [0001] The invention relates to the technical field of natural language processing, in particular to a factual information encoding and evaluation method for generating shipping news summaries. Background technique [0002] In recent years, with the rapid development of information technology, the traditional shipping industry has gradually integrated with artificial intelligence and big data. As an ancient industry, the shipping industry plays an irreplaceable role in the development of human economy. As an important window for the public to understand the development of the shipping industry, shipping news can not only convey various information related to shipping to people, but also have a very important impact on the development of the shipping industry and other related industries. For shipping news texts, if modern technologies such as artificial intelligence are used to extract important information, on the one hand, people can quickly understand the...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a factual information coding and evaluation method for the generation of shipping news summaries, to solve the existing deficiencies, to discover and aggregate factual information of different semantic granularity distributed in different levels of the original text, to realize Accurate and comprehensive representation of textual factual information based on different granularities and contextual relations at different levels, and evaluate and select more important factual information among many factual information, which is beneficial to specific natural language processing tasks such as summary generation, To help the shipping news summary generation model make as few obscure factual errors as possible, and further improve the effect of automatic summary generation

Images