Identifying major tasks and minor tasks within online reviews

Document Type

Article

Source of Publication

Future Generation Computer Systems

Publication Date

9-1-2020

Abstract

© 2017 Elsevier B.V. Many e-commerce websites allow customers to provide reviews that reflect their experiences and opinions about products and services. Such published reviews, whether positive or negative, serve both the consumer and the business. Negative reviews can inform the merchant of issues that, when addressed, may improve the addressed aspect of the business and improve its online reputation. However, when the merchant fails to respond to customers’ concerns, the business faces potential loss of reputation. The Sentiminder system identifies major areas of customer concern, and specific concerns within each area. This helps the merchant to process a large body of reviews and find what needs to be addressed. In this paper we address the problems of quickly finding specific issues and specific comments that are consistently discussed in a negative way. Our technique drills down from the major task areas to more specific issues, assisting the user to accurately determine what issues need attention. The sentiment of reviews on the same topic can vary widely, so we maximize coherence over a variety of six different sentiment assessment techniques. We achieve from about 45% to 65% coherence. These suggestions are implemented in the Sentiminder, an online tool that creates schedules of optimal selections of tasks.

ISSN

0167-739X

Publisher

Elsevier B.V.

Volume

110

First Page

413

Last Page

421

Disciplines

Business | Social and Behavioral Sciences

Keywords

Costs estimation, Document clustering, Online reviews, Sentiment extraction, Social business analytics, Task extraction

Scopus ID

85039157043

Indexed in Scopus

yes

Open Access

no

Link to Full Text

Share

COinS