If you spend any amount of time online, I’m sure that you have seen reviews and comments on Amazon and other places that seem a bit off. Some people simply hop around the web leaving negative comments and reviews of new products out of spite or a desire to harm the company making the product. This is a significant issue for some companies, and Google is backing research at the University of Illinois at Chicago designed to discover fake reviews and comments.
The study hopes to find the organized groups of comment fraudsters and then design an automated process to identify and shut them down. There is no real way to put a dollar amount on the detrimental effects that fake reviews on commerce sites such as Amazon have on products. Harmful fake negative reviews are also common on Yelp and similar sites. At the same time, false positive reviews of the product can also be an issue for consumers.
The researchers have a new algorithm called GSRank that they believe will be able to find fraudulent reviews. The algorithm looks at things such as the time window for fraudulent reviews, expecting that an organized group will post negative reviews within days of each other. The algorithm also looks at deviation because organized groups are all expected to post similar ratings. Content similarity will also be investigated because the groups are expected to copy content between themselves and use the same phrases in different reviews. The algorithm also looks at the size of the group of negative reviews compared to the number of genuine reviews as an indication of spamming.
[via The Register]