ReviewMiner: a novel system for multi-modal review analysis to provide visualized support for decision making

Over the past decades, a number of prominent sites for collecting and presenting reviews in different product categories have seen tremendous growth. However, the growth of number and length of the review text doesn’t provide easier access to knowledge in the review. In fact as the amount of reviews grow, people are less likely to be able to finish reading the helpful reviews. To tackle the information overload situation in review text data, I developed features on top of an existing review analyzer named ReviewMiner. It is a novel system for understanding review data. ReviewMiner is a multi-modal system that combines visualizations of spatial, temporal as well as textual summary of aspects of the review. The highly rich content produced from large amounts of review data empowers the user of the system to make more informed decisions. Because of the underlying LARA algorithm for general aspect rating analysis, ReviewMiner is a framework that can be applied to many categories of review data such as hotel review, restaurant reviews, product review, medical review, and more. In the focused study of hotel review scenario, I demonstrate the system’s major components: (1) natural language search; (2) visualizing the review data with location attributes; (3) analysis of temporal evolution of aspect rating; (4) latent aspect analysis on review text data.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2014

Author dc:creator

• Chen, Gong

Contributors dc:contributor

• Zhai, ChengXiang

• data mining
• review analysis

Statement dc:rights

• Copyright 2014 Gong Chen

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/49401

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/49401