Title: Classification and Analysis of Web Multimedia Data using Principal Component Analysis
Author(s): Siddu P. Algur1, Basavaraj A. Goudannavar2*, Prashant Bhat3
1,2,3 Department of Computer Science,
School of Mathematics and Computing Science,
Rani Channamma University, Belagavi, Karnataka, India
Over the web, the size of multimedia data is increasing in a rapid way. Different types of web multimedia data are used by the web users for different applications. These multimedia data belongs to different categories of domains such as Entertainment, Sports, News and Discussion, Music etc. An automatic classification/prediction of web multimedia data without knowing the content is a challenging and complex research aspect. This paper proposes two approaches to classify web multimedia data viz., classification of web multimedia using dimension reduction technique and classification of multimedia data without reducing the dimensions. To reduce the dimension of the web multimedia metadata, we adopt Principal Component Analysis (PCA) technique to reduce the data dimensions (attributes). The proposed PCA technique involves orthogonal transformation of multimedia metadata values, construction of covariance matrix and computation of Eigen values to reduce the dimensions. The reduced and non-reduced multimedia data are classified separately using DT and KNN classifiers. The classification results of reduced and non-reduced dimensions of multimedia data are analyzed, compared and as a task of KDD.