Google is tackling the growing problem of fake reviews, researching new spam detection software to prevent people ‘gaming the system’ with bogus ratings to drive traffic and sales online.
A new Google-sponsored study, entitled ‘Spotting Fake Reviewer Groups in Consumer Reviews’, by University of Illinois researchers is being used to help to spot opinion spam using a new relation-based algorithm called GSRank (Group Spam Rank).
Fake reviews have long been a bane of the ecommerce industry, and app stores, online retailers and review sites like TripAdvisor are rife with five-star reviews that have clearly been left by people tied to the game in question.
There’s also the issue of for-pay opinion spamming (people being paid to leave favourable reviews).
The new GSRank algorithm looks to catch fraud through a number of key fingerprints from groups of reviews including the time they were posted, the similarity of review/rating, the group size, and when the reviews were posted, because reviews posted too quickly in the life of an app tend to be fake.
The new model of spam detection and has outperformed its competitors in several tests, according to the study.
If successful, the tool could be a huge boon to any app store, though it is likely the Google Play Store would get the benefits since Google backed the research.
Read the full paper here
