ArtiMarker: A Novel Artificially Inflated Video Marking And Characterization Method on YouTube

YouTube is on demand platform, crowd sourced by design and its huge popularity encourages users to adopt fraudulent methods to inflate viewcounts of their videos, thereby increasing the revenue of the video uploader. It is very important to mark artificially inflated videos as it is one of the stepping stone to encourage false information over web and to gain monetary benefits.To address this issue, in this paper we present "ArtiMarker" a novel artificially inflated video marking and characterization (less,medium,high) method on YouTube based on user level feature.The proposed technique described in the paper will help researchers to understand a new efficient method to mark and characterize artificially inflated videos. The proposed parameters in this paper can also be used in identification of spam videos, fake profiles on YouTube. The performance of proposed work is analyzed and demonstrated using various classifier such as Decision Tree, Function-Based and Bayes.The comparison is based on the testing strategy utilized during the experiments;percentage split,cross validation. It has been observed that Bayes network,J48 and Random Forest are best suited for percentage-split while Random Forest is more suited for cross validation schemes.