Anatomy and Empirical Evaluation of an Adaptive Web-Based Information Filtering System

摘要: A case study in adaptive information filtering systems for the Web is presented. The described system comprises two main modules, named HUMOS and WIFS. a user modeling based on stereotypes. It builds maintains long term models of individual Internet users, representing their needs. model structured as frame containing informative words, enhanced with semantic networks. proposed machine learning approach process use an artificial neural network stereotype assignments. WIFS content-based module, capable selecting html/text documents computer science collected from according to interests user. has been created very purpose structure utilized by HUMOS. Currently, this acts interface search engine ALTA VISTATM. An empirical evaluation made experimental settings. experiments focused evaluation, means non-parametric statistics approach, added value terms performance given component; it also usability acceptance system. results are satisfactory support choice model-based Web.