Exploiting Big Data Technology for Opinion Mining

Reviews have an important role in decision making process either for customers or commercial organizations. Hence, it is necessary to develop methods which mine the reviews of customers automatically. This task is referred to Opinion mining or sentiment analysis. Opinion mining covers a wide range of sub-problems in text mining, natural language processing, classification, etc. However with the fast growth of opinion data on the web, the opinion mining process will face some serious problems. Storing, managing and processing such a large volume of data with the traditional approaches are very hard and in some cases impossible. In this paper, we propose two appraches for sentiment analysis of Persian reviews. These approaches are developed based on a Persian sentiment lexicon and a programming language model for distributed systems, called Mapreduce in the Hadoop framework. We examine our proposed appraches with various stations and discussed the effectiveness of the Big Data technology for the opinion mining task.