Automatic Chinese text topic exploration method and system

Abstract

The invention discloses an automatic Chinese text topic exploration method and system; the system comprises a word vector construction module, a text clustering module and a visualization module; theautomatic Chinese text topic exploration method is used in the system, and the problem of long calculation time consumption of a K-Means clustering method can be solved; and more classification feature information is provided, so that text topics can be extracted manually and quickly.

Description

technical field [0001] The invention relates to the field of text theme exploration, in particular to a method and system for automatic Chinese text theme exploration. Background technique [0002] There are many methods of topic exploration, such as topic extraction based on LDA, K-Means text clustering based on unsupervised learning, etc. The LDA topic model is a topic inference based on the perspective of probability and statistics using Bayesian thinking, K-Means clustering The model is a scatter clustering based on the distance of the space vector, which can finally divide the text into different clusters or classes. On this basis, the purpose of text topic extraction is finally achieved through manual further information extraction and induction; under this background , K-Means has the following disadvantages: [0003] 1. The goal of K-Means judgment is to minimize the sum of the squared distances from the cluster members to the actual centroid containing the member. ...

AI Technical Summary

Problems solved by technology

[0003] 1. The goal of K-Means judgment is to minimize the sum of the squared distances from the cluster members to the actual centroid containing the member. As the analyzed data set continues to increase, it is necessary to calculate the distance from all data points to the centroid each time, and the amount of calculation is Continuously increasing, time-consuming increase;

[0004] 2. K-Means can only divide the text into multiple different clusters or classes according to a given number, and does not provide more classification information, which is not convenient for manual and faster induction and extraction of text topics

Images