METHOD AND SERVER FOR EXTRACTING TOPIC AND EVALUATING COMPATIBILITY OF THE EXTRACTED TOPIC

摘要

Provided are a method and a server for extracting a topic and evaluating suitability of the extracted topic. The server for extracting a topic of the present invention comprises: a text pre-treating unit for extracting nouns from a classified document set formed according to classification information, and removing stop words; a keyword extracting unit for extracting keywords which are words representing the classified document set by calculating a weighted value for words with the stop words removed therefrom; a seed selecting unit for selecting seeds which are major words of a cluster obtained by clustering the keywords with related words by calculating a weighted value for the extracted keywords; an initial clustering unit for constituting keywords expressed much in the same sentence as the seeds as one cluster based on the selected seeds; and a cluster combining unit for extracting a classified topic set by combining similar clusters among the constituted clusters.