Effective and Dynamic Multi Domain Keyword Extraction through Word Vectors

  • Neelkanth Poosa, Pranav Sai Marla, M. Venu Gopalachari

Abstract

Online product reviews are a customer's best resource to judge a product. Accurate and useful reviews for a product can be challenging to f ind. Reviews can be tainted, biased, inconsistent or not to the point. This leaves room for consumer-centric review analysis techniques. Multi-Domain Keyword Extraction using Word Vectors is a technique that is aimed at providing the customer with reviews from multiple websites, along with detailed analyses on the reviews, to streamline the customer’s experience. The reviews are dynamically scraped from multiple e-commerce websites using the product’s unique model number. Machine learning is used to accurately identify aspects and key phrases in the reviews, and context-based sentiment analysis is used to establish the average opinion for each keyword. The machine learning algorithms will process the data in the form of word embeddings to accurately f ind the keywords in large texts. Review trustability phase is a unique approach formulated to identify trustable reviews considering various aspects that def ines a review to be credible. Therefore, this method offers e-commerce customers a consolidated review analysis technique.

Published
2020-03-30
How to Cite
Neelkanth Poosa, Pranav Sai Marla, M. Venu Gopalachari. (2020). Effective and Dynamic Multi Domain Keyword Extraction through Word Vectors. International Journal of Advanced Science and Technology, 29(3), 10485 -. Retrieved from http://sersc.org/journals/index.php/IJAST/article/view/27126
Section
Articles