A method and device for automatically sorting sentences

Abstract

The invention provides a method and a device for automatically sorting sentences, wherein the method comprises the following steps: preprocessing a document set with broken sentences to obtain a combined set of sentences; obtaining a word vector dictionary by training the sentence set, and clustering the word vector with the preset synonym word forest; based on a conditional entropy algorithm andthe result of word vector clustering, obtaining the proximity between sentences in the sentence set; adopting a Markov random walk model to sort sentences in the sentence set. The semantic analysis method of the invention can realize the automatic evaluation of the logic of the text sentence, improve the evaluation efficiency, reduce the evaluation error, reduce the influence of the data sparseness, and improve the efficiency of the generation of the sentence sorting result.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a method and device for automatically sorting sentences. Background technique [0002] With the rapid development of Internet technology, the research on automatic grading of Chinese composition is gradually rising. It is of great significance to improve the efficiency of composition grading, fundamentally eliminate the inconsistency of composition evaluation, and control the error of grading. Due to the complexity of Chinese language logic, most of the existing research evaluates composition from the perspectives of vocabulary usage, grammar, expression, composition length, use of associated words, rhetorical techniques, and the consistency of the topic of the article. Evaluation of internal logical rationality. However, in composition evaluation, logical rationality is also an important indicator for evaluating language use ability. The reasonable logic between sent...

AI Technical Summary

Problems solved by technology

[0004] However, there are still many problems in the above studies. For the first and second types of research, the main problems are: the use of time information, the inheritance relationship between sentences, the theme of sentences, etc., has relatively large limitations, and it is impossible to use these specific methods. Sentence sorting of the text of the information; in addition, due to the lack of understanding of natural language by machines, it is also a major difficulty to rely on subject words, time words, hidden time recognition, and invisible associated words mining

For the third type of research, its shortcomings are mainly: relying on large-scale corpus to calculate the word collocation between sentence pairs, the parameter space is large, and the problem of data sparseness is prone to occur, which is not conducive to the subsequent calculation of proximity

Images