Network public opinion topic feature extraction method and system

Abstract

The invention provides a network public opinion topic feature extraction method and system, which firstly consider the difference between general network public opinions and large-scale network public opinions, the method is suitable for topic feature extraction of the large-scale network public opinions; secondly, a method based on a co-word network is improved, a method of combining word frequency with ITF / PDF is adopted during keyword extraction, and the weights of the keywords can be described more accurately; thirdly, the non-threshold inter-stage evolution network method adopted by the invention is simple, can retain subtle association between topics, and accords with topic evolution logic; finally, the invention provides an event-driven topic spatio-temporal feature discovery method, and address tags of the text corpora are replaced by address tags of events, so that the situation that the text corpora have no address tags can be matched.

Description

technical field [0001] The invention relates to the technical field of data analysis, in particular to a method and system for extracting network public opinion topic features. Background technique [0002] At present, public opinion topic feature extraction is mainly divided into two steps, the first is topic identification and extraction, and the second is topic feature discovery. For the first step of topic identification and extraction, currently, commonly used methods mainly include clustering-based methods, topic model-based methods, and co-word network-based methods; for the second step of topic feature discovery, the content can include two points. One is spatio-temporal features, and the other is evolutionary features. The current topic spatio-temporal feature discovery method is mainly based on the text corpus with address tags, and the address tags are counted and spatio-temporal mapping; the method usually used in the topic evolution feature detection is based on...

AI Technical Summary

Problems solved by technology

[0004] First of all, the existing public opinion topic feature extraction methods do not distinguish between general public opinion and large-scale public opinion. There are huge differences in the time span, secondary events, and corpus size of the two kinds of public opinion.

Secondly, in the process of topic identification and extraction, the existing methods have the following problems. The clustering-based method is random and will introduce interference information to affect the accuracy; the topic model-based method needs to determine the number of topics in advance, but large-scale network Public opinion spans a large span and there are many topics, and the topic model method will cause the problem of missing topics; although the method based on the co-word network can present a scientific cognitive structure, there are still areas for improvement in specific steps. For example, most methods use word frequency Or subjective judgment to extract keywords, this keyword extraction method is subjective and the description of keyword weight is not accurate enough

Thirdly, the current spatio-temporal feature discovery method for topics is not suitable for data without address labels, and the existing methods reflect the hot spots of posting or comments, rather than the concerns of netizens.

Finally, the existing topic evolution detection method uses a co-word network method with a threshold. Through the setting of the threshold, some topics are filtered, so topic relationships with less relevance cannot be retained.

Images