Dr Neeru, Vishal Shrivastava


Today, most of the products and services are available online and the vendors are directly connected with the customer to deliver these products and services. While providing these product and services, the vendors also take the feedback or opinions or the reviews about the product or services. These opinions are considered as the customer interest or the satisfaction. One of such online review systems is adapted by mobile industry to identify the user interest in mobile. A mobile is described different aspects such as camera quality, built quality, os etc. As the user review is accepted in the textual form, the analysis is required on these reviews to adapt the valuable information from it. This valuable information is divided in two main categories called the mobile aspect identification and the user sentiments identification. The sentiments of user are considered as quality measure of mobile as well as mobile aspects. These sentiments are defined by specific positive and negative adjectives used by the customer or reviewer. Based on these reviews, the mobile quality is analyzed and it gives the recommendation to other users to use that mobile or not. If the review is positive the sale of  that mobile will be increased but if the review is negative, the sale of that mobile will be decreased. Because of this there is a requirement to analyze these reviews effectively.

In this proposed work, a review dataset is considered to perform the sentiment analysis. This database is having the multiple reviews associated with a particular mobile. In this proposed work, a weighted approach is defined for sentiment analysis. Different adjectives used by the customer are assigned by different weightage based on the adjective criticality. Now each review associated with the mobile is analyzed under the aspect and adjectives. The aggregative weighted analysis on the mobile reviews is performed to obtain the overall sentiment associated with the mobile and the mobile quality will be identified.


Naïve Bayes, Aspect Based Sentiment Analysis, Mobile Review, WordNet, Aspect Level Opinion Mining

Full Text:



  • There are currently no refbacks.