Granular Computing Models in the Classification of Web Content Data

The paper addresses two problems of web content mining, such as scene-region classification (applicable to image annotation), and image based spam detection. To solve these problems, we describe two granular computing (i.e., with rough-fuzzy and rough-wavelet granular spaces) based pattern classification models. These models can be used to design intelligent agents which may provide an improved solution to web mining. Neighborhood rough sets are used in the selection of a subset of these granulated features of models. Both the models explore mutually the advantages of fuzzy/wavelet granulation and neighborhood rough sets. The superiority of these models to other similar methods is established with various performance measures.