A comparative method for different aspect based products features in online reviews of different languages
Manushree A. M, M J Adarsh, Pushpa Ravi Kumar · 2017
User review is the most valuable data. This review contains data in the form of opinion about a particular product or entity in detail. Using reviews, conclusion about any product can be drawn and also, user related complaints while using the product or expectation of users from the product manufacturer can be identified clearly. These reviews may also serve as a feedback to the product manufacturer to improve or to correct their flaws regarding that product. The challenge is to manage and analyze the huge sets of these reviews in a convenient way. While analyzing and managing reviews there exists a set of problems such as words, which are misspelled, grammatically incorrect sentences and also reviews written in languages other than English. The analyzer cannot expect the user to write reviews in a way as he/she expects. The current trendy generation believes the use of short forms, misspelled words as a current trend. Therefore, it is important to deal with these trendy problems as well. This paper proposes a method that put forwards an idea to deal with problem related to misspelled words in a product review and also an attempt to deal with multiple languages at a time and a comparison between SentiWordNet and TextBlob has been made to show the difference in accuracy while computation. The results are obtained based on the polarity score. The polarity score for each sentence in the review is assigned using the TextBlob. This method also put forwards a technique, which can overcome most commonly committed spell mistakes by the reviewer in context of needed aspects, also a solution to deal with reviews written in other languages.