ROUGH SET THEORY APPLICATIONS ON MEASURING TEXT MINING TASKS

Mining databases are becoming more essential nowadays. Extracting knowledge by data mining process is facing some issues when its prediction expecting the level of the perfection or accuracy. Recent years, Rough set theory (RST) has been accepted as an effective technique to discover hidden patterns in data and also it's is known for its simplicity. Its concept of attribute reduction executed using approximations that have been used in many places of mining. One of the main issues in data mining is dimension reduction where researchers proposed many methods. RST, the efficient approach simplifies some problems when other processes make in text mining tasks. In text mining, the primary process is preprocessing where we need to apply some filters to remove the irrelevant words and form such kind words using properties that makes classifier uncomplicated. In this paper, we are mining the database of product reviews using Rough set reduction concept by testing the created models combinations of ngrams that identify the word dependency of RST in text mining. Experiment results confirm that combination of unigrams and bigrams to perform well in comparison with other models.