The Application Research of Topic Word List In Text Automatic Classification

When the traditional text classification technologies classify academic dissertations, the dimension of extracted feature terms is high, and they can´t represent the theme of thesis. it makes the efficiency is very low and the accuracy rate is not high. The topic words are small in quantity and can reflect the theme of thesis well. Accordingly, the paper proposes to extract the topic words with topic word list and uses topic words as feature terms. Then using the Bayesian classification method classifies vast texts. The experiments show that the Bayesian classification method using topic words as feature terms can greatly reduce the dimension and improve the efficiency of classification, when the dimension of feature terms is equivalent, the accuracy of Bayesian classification method using topic words as feature terms is also higher than the traditional Bayesian text classification methods.