Method for detecting burst topic in user generation text stream based on graph clustering

Abstract

The invention relates to a method for detecting a burst topic in a user generation text stream based on graph clustering and belongs to the technical field of internet data mining. By the method, a graph-based new field of view relative to the conventional topic detection problem is provided, and the detection problem of the burst topic in the text stream is converted into a typical graph clustering problem, so the problem can be solved by using the conventional graph theory method. The method comprises the following main steps of: acquiring the text stream; detecting the burse topic; constructing a burst word graph; and clustering burst words. The method aims at the detection of the burst topic in the user generation text stream and has the performance which is superior to that of the conventional method based on document clustering, a probability topic model and burst characteristic clustering.

Description

technical field [0001] The invention belongs to the technical field of Internet data mining, in particular to a method for detecting sudden topics in text streams. Background technique [0002] Accurate detection of emergent topics from massive user-generated texts is of great significance for government decision-making and commercial promotion. A breaking topic can be a popular event that happens at any moment, or it can be a network activity spontaneously and widely responded to by bloggers within a period of time. Events can be unpredictable, like a volcanic eruption, or predictable, like a presidential election. Activities are generally unpredictable, such as the spread of blog quizzes (Internet quiz). A sudden topic often has a short duration and is heatedly discussed by a large number of netizens. [0003] However, due to limitations such as weak modeling and inflexible parameter settings, existing text clustering, probabilistic topic models, and bursty feature extr...

AI Technical Summary

Problems solved by technology

However, the community structure detection method is still insufficient to solve the clustering problem of sudden words.

First of all, this method uses the co-occurrence number of two graph vertices to measure the correlation between vertices, and this non-normalized measure cannot distinguish between the same topic burst words and Linking Edges Between Burst Words on Different Topics

Secondly, this method is not suitable for the weighted graph structure with the characteristics of "fully connected", because the betweenness measure degenerates in the fully connected graph, and it cannot distinguish between the words of the same topic burst and the words of different topic bursts. even side

Images