NetSpam: a Network-based Spam Detection Framework for Reviews in Online Social Media (Jul-2017)
The main aim of this project is to detect spam reviews from online social media using user based and review based detection scheme.
In our proposed system we propose to determine the relative importance of each feature and show how effective each of features are in identifying spam from normal reviews and also improves the accuracy.
Early Time Frame: Spammers, usually write their spam reviews in short period of time for two reasons first, because they want to impact readers and other users, and second because they are temporal users, they have to write as much as reviews they can in short time. Spammers try to write their reviews, in order to keep their review in the top reviews which other users visit them sooner. To avoid this type of spam reviews we can calculate the days between last and first review of the particular user to detect spam reviews.
Content Similarity: Spammers, often write their reviews with same template and they prefer not to waste their time to write an original review. In result, they have similar reviews. To avoid this each and every user review will be compared with spam review templates to detect spam reviews.