Abstract

The product reviews in the fields of camera and mobile phone of online shopping sites are viewed as research corpus for opinion mining, the paper proposes an extraction algorithm of domain evaluation object based on statistics and grammar rules, considers the relative frequency of evaluation object, regulates weighing of evaluation words according to document frequency and the co-occurrence frequency between topic words and the domain authoritative words, uses the improved Term Frequency-Inverse Document Frequency (TF-IDF) method to extract topic words, then utilizes the t-support value to filter out the redundant words and gets the final product reviews topic words. The experimental results show that the precision rate is 83.77%, the recall rate is 84.37%, which indicate the efficiency and feasibility of the proposed method of topic mining on product reviews.

Paper Details