Recently, discovering outliers among large scale web videos have attracted attention of many web mining researchers. There are number of outlier/abnormal videos exists in each category of web videos such as- ‘Entertainment’, ‘Sports’, ‘News and Politics’, etc. The task of identifying and manipulate (to remove from the web or to share with others in the web, or to watch/download from the web etc) such outlier web videos have gained significant important research aspect in the area of Web Mining Research. In this work, we propose novel methods to detect outliers from the web videos based on their metadata objects. Large scale web video metadata objects such as- length, view counts, numbers of comments, rating information are considered for outliers’ detection process. The outlier detection method –Local Outlier Factor (LOF) with different nearest neighbor values (with K=3, K=5 and K=7) are used to find abnormal/outlier web videos of same age. The resultant outliers are analyzed and compared as a step in the process of knowledge discovery.

Outliers, Lcal Outlier Factors, Inter-Quartile Range, Web Video Outliers, Clustering, YouTube

[1] Chueh-Wei Chang, Ti-Hua Yang and Yu-Yu Tsao, “Abnormal Spatial Event Detection and Video Content Searching in a Multi-Camera Surveillance System”, MVA2009 IAPR Conference on Machine Vision Applications, May 20-22, 2009, Yokohama, JAPAN.
[2] Fan Jiang, Ying Wu, Aggelos K. Katsaggelos, “Abnormal Event Detection from Surveillance Video by Dynamic Hierarchical Clustering”, Northwestern University, USA.
[3] Tushar Sandhan et al., “Unsupervised learning approach for abnormal event detection in surveillance video by revealing infrequent patterns”, IEEE 28th International Conference on Image and Vision Computing, 2013- New Zealand
[4] Thi-Lan Le and Thanh-Hai Tran, “Real-Time Abnormal Events Detection Combining Motion Templates and Object Localization”, Advances in Intelligent Systems and Computing 341, DOI 10.1007/978-3-319-14633-1_2, Springer International Publishing-2015, Switzerland.
[5] Yang Cong et al., “Abnormal Event Detection in Crowded Scenes using Sparse Representation”, Pattern Recognition, January 30, 2013
[6] Cewu Lu et al., “Abnormal Event Detection at 150 FPS in MATLAB”, The Chinese University of Hong Kong.
[7] Yang Cong et al., “Sparse Reconstruction Cost for Abnormal Event Detection”.
[8] Bin Zhao et al., “Online Detection of Unusual Events in Videos via Dynamic Sparse Coding”, 2011.
[9] Mahmoudi Sidi Ahmed et al., “Detection of Abnormal Motions in Video”, Chania ICMI-MIAUCE’08 workshop, Crete, Greece, 2008.
[10] Du Tran et al., “Video Event Detection: From Subvolume Localization to Spatiotemporal Path Search”, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 36, 2014.
[11] Dataset for "Statistics and Social Network of YouTube Videos", http://netsg.cs.sfu.ca/youtubedata/.
[12] Siddu P Algur, Prashant Bhat, "Metadata Based Classification and Analysis of Large Scale Web Videos", International Journal of Emerging Trends & Technology in Computer Science (IJETTCS) , Volume 4, Issue 3, May - June 2015 , pp. 111-120 , ISSN 2278-6856.
[13] Siddu P. Algur, Prashant Bhat, Suraj Jain, “The Role of Metadata in Web Video Mining: Issues and Perspectives”, International Journal of Engineering Sciences & Research Technology, Volume 4, Issue 2, February-2015.
[14] Chirag Shah, Charles File, “Infoextractor – A Tool for Social Media Data Mining”, JITP 2011.