期刊名称：IAENG International Journal of Computer Science
印刷版ISSN：1819-656X
电子版ISSN：1819-9224
出版年度：2018
卷号：45
期号：4
页码：560-568
出版社：IAENG - International Association of Engineers
摘要：In the present paper, applying a text analysis of customer feedback received from online stores reviews in Bulgarian is considered. For this purpose, a freely available dataset with customer reviews in Bulgarian is created. Besides, an approach to automatically association of the reviews with predefined categories reflecting the customer satisfaction is proposed. An enrichment of the vector space model for document presentations based on computing a pointwise mutual information measure of the terms in regard to the categories is performed. Experiments are conducted on the collected dataset by using different classifiers. The results of measures to assess the classification performance are provided.
关键词：text classification; customer reviews; pointwise mutual information
