Subject extract method based on word simultaneous occurences frequency

Abstract

The invention is a theme extracting method based on word coexisting frequency, which belongs to information processing field. The invention uses word as base processing unit, at first, accounts the times of each word to the inputted text, deletes the word occurring once and the normal words, then, accounts the coexisting frequency of word in the text, and acquires the information quantity of coexisting between word and word, the static result is reserved in the matrix, then calculates the information in each sentence and segment because of the coexisting relation of word in each sentence or segment, finally carries on the arrangement and output of the theme sentence and segment after the weight adjustment, thus realizes the extraction of theme sentence or segment.

Description

technical field [0001] The invention relates to a method for automatically extracting text topics, in particular to a method for extracting topics based on word co-occurrence frequency. The field of information processing technology for networks. Background technique [0002] Topic extraction is one of the basic tasks of automatic text processing. Topic extraction can be performed at multiple levels such as topic words, topic concepts, topic sentences, and topic paragraphs. The topic extraction step usually applies various weighting algorithms to calculate the contribution of topic words, sentences, paragraphs, etc. to the text topic, and select the topic words, sentences, paragraphs, etc. that contribute greatly. However, the weighting and extraction algorithms are mostly statistical and empirical weighting systems, which do not take into account the relationship between words appearing in the text, especially when the text style changes, the empirical and statistical wei...

Images