Using Appearance Information for Web Information Credibility Analysis

We have been developing a Web information analysis system named WISDOM, which supports human judgment of information credibility by analyzing Web pages based on various criteria such as the sender, content, and appearance. This paper describes use of appearance information on WISDOM. Appearance information means document style and superficial characteristic, such as page layout, sentential style, the number of advertisement, and whether or not there is contract address. We use appearance information for not only Web information credibility analysis but also Web page classification. This paper also reports the extraction method of appearance information, implementation on WISDOM, and results of preliminary experiment.