A Classification Method of Enterprise Industry

Abstract

The invention discloses an enterprise industry classification method. The method utilizes the graph splitting clustering algorithm of semi-supervised learning to effectively extract the main business keywords of the enterprise, and uses the extracted keywords as features based on the gradient lifting decision tree to train The cascade classifier classifies enterprises by industry, which solves the cumbersome problem of manual classification. The specific method is: 1) Use the word vector and semi-supervised graph splitting clustering algorithm to extract the main business keywords of the enterprise, eliminate the junk words, and construct the keyword library; 2) Input the extracted keywords as features to train the cascade classifier , the classifier at each level classifies the enterprises, and the unclassified enterprises are classified by the classifier at the next level. The invention not only can automatically construct, update and classify keywords, but also solves the problem of classifying industries of tens of millions of enterprises, and can effectively solve the problem of manual labeling.

Description

technical field [0001] The present invention relates to the research field of data classification methods, and more specifically, relates to the extraction of industry keywords. In the case that the business scope of the enterprise overlaps with multiple industry descriptions, the fusion of semi-supervised graph splitting and clustering and cascading gradient boosting decision trees The enterprise industry classification method. Background technique [0002] In the industry classification standard issued by the National Bureau of Statistics of the People's Republic of China in 2013, it is divided into 20 first-level industries and 96 second-level industries. The industry label of an enterprise is an important field, and there are tens of millions of enterprises across the country, and many enterprises are incubated every day. How to quickly classify enterprises by industry is an important issue. In the previous industry classification norms, the industry to which an enterpr...

AI Technical Summary

Problems solved by technology

In addition to this, a single classifier model relies too much on the coverage of sample descriptions, and is less accurate when classifying a new sample of a description that has never appeared before.

Images