The popularity and use of e-commerce are increasing day by day. Recent trends have shown that many people are now buying their products online through different e-commerce sites such as Flipkart, Amazon, Snapdeal, etc. Customers review and rate the products they have bought over multiple independent review sites such as gsmarena, social networking sites like Facebook, blogs, etc. Customers can also comment on the quality of service they have received after buying their product. These online reviews are of immense help to potential buyers for that product in decision making and also to manufacturers/sellers to get an immediate feedback about the product quality, product performance, after sales service, etc. As the number of reviews for a product is usually large, it is next to impossible to go through all the reviews and form an unbiased opinion about the product. Also, there are multiple sources of these online reviews. Hence, online review mining is gaining importance.
A product can have many product features, wherein some features are more important than others. Review usefulness can also be increased by ranking the product aspects as per their importance and popularity. Ranking product aspects manually is very difficult since a product may have hundreds of features. So, an automated method to do this is needed.
This paper presents a methodology for co-extracting opinion targets and corresponding opinion words from online opinion reviews as well as for product aspect ranking.
Title = "Effective Co-Extraction of Online Opinion Review Target-Word Pairs and Product Aspect Ranking",
Journal ="International Journal of Computer Applications Technology and Research(IJCATR)",
Volume = "6",
Pages ="349 - 416",
Year = "2017",
Authors ="Chandana Oak , Prof. Manisha Patil"}
The paper proposes a method for co-extracting opinion targets and words from reviews
Partial supervision technique is used for data processing
Opinion TW Co-extraction Algorithm is proposed for co-extraction of opinion target-word pairs
Product Aspect Ranking is also proposed to rank product features as per their importance.