Concept-Level Sentiment Analysis of Online Hotel Reviews

Xiaonan Jing, Purdue University

Abstract

With the growing number of online communities, more and more people are sharing their product reviews or experiences on these platforms to express their opinions. These reviews help companies understand what the customers prefer and what improvements can be made in the future and help many new customers make decisions based on these reviews. The hotel industry is one of the domains where online reviews most influence the choices of new customers. In order to study what emotions are being conveyed to the reader, sentiment analysis is often performed to discover the attitudes of reviews. However, it is difficult to determine the overall sentiment of a review since it may contain both positive and negative sentiments towards different aspects at the same time. Therefore, sentiment analysis is performed at the concept level to understand the precise meanings in reviews. This thesis introduces an approach for discovering the sentiment features from online hotel reviews using a Word2Vec model combined with the spherical k-means clustering algorithm. The Word2Vec model generated the semantic features for the clustering process and many of the resulting clusters were recognized as meaningful by annotators. Additionally, within the clusters that were labeled to have the same sentiment feature, different semantic meanings across these clusters were observed. The consistency of the clusters was confirmed using a “split value” heuristic which was computed based on cluster similarity and cluster diversity. The split value also served as an indicator for how much ambiguity the dataset contained which helped analyze the resulting clusters.

Degree

M.S.

Advisors

Rayz, Purdue University.

Subject Area

Information Technology|Information science

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS