Filtering Web Pages with Application Ontologies

Automatically recognizing which HTML documents on the Web contain objects that are “of interest” for a user is non-trivial. As a step toward solving this problem for information filtering in which a user expresses a long-term need for information with a specific profile, we propose an approach based on ontological descriptions. The HTML documents we consider include semistructured HTML documents, HTML tables, and HTML forms. Given the keywords and values and kinds of values recognized by an ontological specification in an HTML document, we apply several heuristics: (1) a density heuristic that measures the percent of the document that appears to apply to the application ontology, (2) an expected-value heuristic that compares the number and kind of values found in the document to the number and kind expected by the application ontology, and (3) a grouping heuristic that considers whether the values of the document appear to be grouped as application-ontology records. Then, based on machinelearned rules over these heuristic measurements, we determine whether an HTML document contains objects of interest with respect to an application ontology. Our experimental results show that we have been able to achieve about 95% for both recall and precision.