Title:	THE PYTHIA WEB FILTERING AGENT
Author(s):	Spyros Vrettos , Lambros Katis , Andreas Stafylopatis
ISBN:	972-98947-3-6
Editors:	Nuno Guimarães and Pedro Isaías
Year:	2004
Edition:	Single
Keywords:	Web Filtering, Intelligent Agents, Machine Learning.
Type:	Full Paper
First Page:	1484
Last Page:	1491
Language:	English
Cover:
Full Contents:	click to dowload
Paper Abstract:	Web Directories constitute the primary form of organizing the wealth of information published on the web. The corresponding hierarchical tree-like structure defines a Taxonomy which can be used to assign metadata to the web pages that belong to the leaf nodes of the structure. Additionally, every web page in the directory is described by a short description to characterize its contents. Therefore, a query applied to a Web Directory retrieves a result list, quite rich of metadata (directory path and short description) for every page in it. In addition to large web directories and portals, that are intended to organize a respectable portion of the web, Taxonomy building is also frequently used in organizing user’s interests in the form of bookmarks kept in browsers. Machine learning (Text Classification) is used to automate the assignment of a web page to a leaf of a Taxonomy. However, creating bookmark classifiers is a challenging task because few training samples are given. Pythia is a web filtering agent that induces term weighting according to the metadata on the search results, thus improving classification according to the user’s bookmarks. Web filtering is performed through an architecture that uses a Mediator Agent between the User Agent and the web.
	Go Back