SENTIMENT REFINEMENT BY EXTRACTION OF HIDDEN INFORMATION FROM CUSTOMER COMMENTS

Abstract

An important area that leads to failure or success of a product lies in customer comments posted online at various social media sources. What customer will buy will vary as per pattern of what other old customers write in comments that are available online. These reviews can have different emotional nature varying from positive to negative and sometimes neutral. There are various elements and off-shelf tools that can help us to classify these posts. However, there can be various loopholes that can modify the real emotion of these comments because of Unidentified/Hidden information in neutral comments”, “Wrong keyword extraction while splitting words”, “fake comments based on frequency of duplicate comment or reviewer”. This report is use to evaluate and categorize user feedback reported on Amazon platform for e-commerce and proposes solution in the form of a flow chart and algorithm. Data may be of many types but as per the globalization of models and techniques nowadays clustering on uncertain data have become one of the essential tasks in mining uncertain data. Earlier approaches faced the various roadblocks in modeling similarity between uncertain areas and bringing up an efficient approach of computation method. The efficient retrieval of meaningful information from the databases has always been remained as a significant issue in Data Mining.