User clustering and short text clustering method based on social network short text stream

Abstract

The invention aims at solving the problem that social contact factors, 'semantic drifting' and sparsity of a short text are not given into considerations in a user clustering and short text clustering method based on semantics. The invention provides a user clustering and short text clustering method based on social network short text stream topic modeling. The method comprises the first step of acquiring a corpus; the second step of conducting pretreatment on the corpus; the third step of conducting topic modeling based on the short text data stream in the social network; the fourth step of deduction and sampling; the fifth step of conducting clustering on users; the sixth step of conducting clustering on short texts. According to the user clustering and short text clustering method based on social network short text stream topic modeling, the three factors of 'semantic drifting', 'sparsity of the short text', and 'the social network' which can influence the topic modeling are fully considered, the problem that analyzing clustering of the user and text through the social network short text stream is lack of social semantic information is solved, and thus precision of an existing clustering algorithm is sharply improved.

Description

technical field [0001] The present invention relates to the field of computer technology, and more specifically, to a method and system for clustering users and short texts based on social network short text streams. Background technique [0002] With the popularization of mobile Internet and the rapid development of social networks, hundreds of millions of user data have been deposited on social networks, how to analyze the short texts published by these users for user clustering and short text clustering has become a very important topic. However, the existing methods do not have an effective method for performing user dynamic clustering on the latent semantic information of short text data streams in social networks, so the present invention proposes an effective dynamic clustering method to solve the problem of latent semantic clustering of short text data streams in social networks. Analyze the problem of user clustering and short text clustering. [0003] The present ...

AI Technical Summary

Problems solved by technology

(But the LDA that this invention relates to is not suitable for the short text data stream in the social network, there are three reasons, 1. Time factor is not considered

2. Social factors are not considered 3. User expression habits are not considered

Therefore, it is impossible to truly capture the subject characteristics of users in social networks

Images