A Knowledge based Approach to Analyze the Sentiment of Online Reviews

In this paper, we are going to present a work which resolves the sentiment of the online reviews. In this work, first the online reviews about several entities are collected from the internet using a crawler program. Next, the sentiment of individual review is resolved using two algorithms. First one is the Lexical Similarity based approach and second algorithm is based on Context Expansion strategy. Contexts of the reviews are expanded with the help of English WordNet. The algorithms are tested on online reviews of 15 entities from 5 different domains. The lexical similarity based algorithm produced 62% accuracy w.r.t. human judgment and context expansion based algorithm produced 82% accuracy w.r.t. human judgment. The challenges which are faced during the experiment are discussed in the pre¬conclusion section.