Web Rec Final Report

This document summarizes a web recommender project. It describes three algorithms - baseline, HTML structure-based, and semantics-based - for recommending relevant web pages based on an input page. An evaluation of the algorithms on five topics found the structure-based algorithm performed best overall, particularly for the topic of "entropy". Error analysis revealed ways to improve the semantic algorithm, such as combining title words with named entities. The project demonstrated the importance of topic on recommendation success and provided insights into developing effective web recommending tools.