Automatically evaluating of product comments with text mining techniques

The one of the important factors which affect consumers? purchasing behavior is known comments of another consumers. So purchases made in internet media make it easy to reach these views of them about products or services. However this advantage complicates to analyze all comments because this media brings about increasing the number of comments at the same time.In this study, summary knowledge is aimed to extract from comments so as to save time for consumer. Consumer reviews of the selected product were primarily processed into morphological analysis by getting them on the web sites called www.hepsiburada.com. Word types and prefix or suffix was determined as a result of analysis of these words. Also, words and adjectives characterizing them were identified in order to extract knowledge indicating negative and positive meanings from these texts written in natural language.A software system was developed to evaluate results with aim of determining the desired characteristics by creating tree structure. This system coded by using depth-first search algorithm. It was implemented in JAVA language being fully object-oriented programming language and having a wide range of documentation. Cause of selecting this language was that Zemberek being natural language processing library is built up by using it. NetBeans was chosen as this software developed in JAVA codes. Data being result of this process was stored in SQL database. When this data is queried according to the desired structure, numerical information designating the degree of consumers? satisfaction was obtained from comments of them about product characteristic.