The Internet is slowly shaping to be the primary information source that fulfils all the needs of a person. Whenever someone plans to buy a product, they tend to consult with the reviews online to get a clear idea of the product in terms of its various aspects. The problem is that the information available about a single product is so much in volume that the users not be able to extract the information they require from this massive amount of data. The paper proposes a system that generates a temporal aspect based text summary of user opinions that are collected from different sources across the Internet with their time-stamp. These comments are broken into sentences and sub-sentences after predefined based classification. Then, Sentiment analysis is performed. The time relationship is taken into account, and the causal relationship is identified at the deflection points or the time frames during which there is a significant opinion change. The major advantage of this system is that the changes in user opinions with time can be traced and the cause of this sentiment change can be found out in addition to offering customers a quick, convenient and easy way to consume information about a product to help them decide whether or not to purchase it. It also helps enterprises to get relevant insights related to their products based on the customer reviews online.
Authors: Pankaj Dadheech, R. Sheeba, R Vidya, P. Rajarajeswari, Padmini Srinivasan, C. Sathiya Kumar, Sudhakar Sengan
DOI: https://doi.org/10.1166/jctn.2020.9426
Publish Year: 2020
