Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
State-of-art approaches for review spammer detection: a survey
Dewang R., Singh A. Journal of Intelligent Information Systems50 (2):231-264,2018.Type:Article
Date Reviewed: Jun 27 2018

The constant growth of e-commerce and online markets presents new paradigms for buyer and seller behavior. Most of us searching for goods or services online ask for information sources containing not only a description of the service/product, but also other users’ opinions and valuable reviews. In this process, many social networking services and forums provide reviews and user opinions of products and services. Trust is the most important factor influencing any kind of business; this is also true for e-commerce, where trustworthiness is hard to fully realize due to the virtual nature of e-business processes. Thus, many vendors include review/opinion sections (user-generated content) to help customers select goods/services according to their wants and needs.

Unfortunately, such generated information lacks quality control because anyone can write anything, sometimes resulting in low-quality reviews and fake reviews (review spam) that may misinform customers and affect their final business decisions. These fake reviews are often oriented to the product/service itself, the brand, or to the operational environment of the entire product/service. Further, spam reviews are written by spammers to either advertise the product features or defame them. Discovering and differentiating fake reviews is a hard task, though research in this field is beginning to find affordable methods and techniques to discover such behavior. Many techniques and methods are in use today, for example, the factor graph model, the GSRank method, the author spamicity model, and the longest common subsequence algorithm.

Several surveys of these methods and techniques exist. However, this study is essentially different in approach and presentation, making it a very attractive resource for further research in the field. Dewang and Singh provide insight into spam, its origin, and the history of spam forms within various online services in order to combine various spammer detection techniques. This process undoubtedly includes machine learning techniques. Besides presenting group and individual spammer detection models on the basis of review datasets used in existing work, the authors extend the study to include features used in group and individual spammer detection. Such an approach gives a more visual and informational look at review spammer detection techniques, which the authors divide into supervised learning techniques (for example, graph-based techniques) and unsupervised techniques, in which the Bayesian network model and clustering are each described in a clear and concise way.

It is noteworthy that although synthetic, artificially generated datasets are sometimes used in review spam detection, most of the current research is based on real-world datasets; the authors understand the importance of this, focusing on the representation of review datasets. These datasets are mostly unstructured texts, and the authors demonstrate the need for a vector space model (that is, a simplifying representation used in natural language processing known as “bag-of-words”). The authors also evaluate hybrid methods, rank learning approaches, and an activity model. Such a presentation gives extensive insight into state-of-the-art spammer detection techniques, making this survey highly recommended reading not only for students in e-commerce, but also for professionals dealing with reviews of products and services, whether they are service providers or online market researchers.

Reviewer:  F. J. Ruzic Review #: CR146115 (1810-0548)
Bookmark and Share
  Featured Reviewer  
 
Electronic Commerce (K.4.4 )
 
 
Security (K.4.4 ... )
 
Would you recommend this review?
yes
no
Other reviews under "Electronic Commerce": Date
Do electronic marketplaces lower the price of goods?
Lee H. Communications of the ACM 41(1): 73-80, 1998. Type: Article
Feb 1 1999
On-line profits
Keen P., Ballance C., Harvard Business School Press, Boston, MA, 1997. Type: Book (9780875848211)
May 1 1998
Secure electronic transactions
Loeb L., Artech House, Inc., Norwood, MA, 1998. Type: Book (9780890069929)
Jul 1 1998
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy