Comparing title only and full text indexing to classify web pages into bookmark categories

Abstract

Web browser bookmark files are used to retain and organise records of web sites that the user would like to revisit. However, bookmark files tend to be under-utilised, as time and effort is needed to keep them organised. We use two methods to index and automatically classify documents referred to in 80 bookmark files, based on document title-only and full-text indexing, respectively. We evaluate the indexing methods by selecting a bookmark entry to classify from a bookmark file, and recreating the bookmark file so that it contains only entries created before the selected bookmark entry. Classification based on full-text indexing generally outperforms that based on document title only indexing. The ability to recommend the correct category at rank 1 using full-text indexing ranges from 20% to 41%, depending on the number of category members. However, combining the approaches results in a increase to 37% — 59%, but we would need to recommend up to two categories to users. By recommending up to 10 categories, this increases to 58% — 80%.