Method and system for automatically generating article based on description text

Abstract

The invention discloses a method and a system for automatically generating an article based on a description text. The method comprises the following steps: receiving a description text of a to-be-generated article; determining a theme of the description text, and taking the theme as the theme of the to-be-generated article; constructing a thematic word graph according to a text corpus corresponding to the theme; and generating an article corresponding to the theme according to the constructed thematic word graph. The method and the system which are disclosed by the invention have the advantage that article generating efficiency and accuracy can be improved.

Description

technical field [0001] The invention relates to the field of natural language processing, in particular to a method and system for automatically generating articles based on description text. Background technique [0002] In recent years, with the rapid development of artificial intelligence, more and more manual work has been replaced by machines. For example, the machine automatically grades the papers. When marking the papers, give the machine some prior knowledge in advance, and the machine can give each test paper a fair and just result. Scoring, the prior knowledge, such as the scoring standards given by domain experts or the standard answers of the test papers, etc.; another example is the automatic scoring of the oral test by the machine. When scoring, give the machine some prior knowledge in advance, such as the reference answers for each test question Wait. The realization of automatic marking by machines and automatic marking of oral exams by machines has greatly...

AI Technical Summary

Problems solved by technology

The traditional method of generating reference articles generally manually finds articles with the same subject as the description text according to the description text of the article to be generated. If an article with the corresponding topic cannot be found, it needs to be manually written, and the manual workload is relatively large; Especially for exams that are automatically graded by machines, an exam often includes multiple sets of test papers, and each set of test papers requires a large number of reference articles, which are used as reference answers for the test composition and grade the students' composition; Reference articles require a lot of work. At the same time, when searching or writing reference articles manually, there is a certain degree of subjectivity when they are used as reference answers. For the same set of test questions, when different people search or write reference articles, the topics will be different, which affects reference. Consistency of answers reduces the accuracy of machine scoring

Images