Semantic Document Clustering on Named Entity Features

TL;DR

This paper enhances document clustering by incorporating named entity features into the vector space model, improving semantic understanding and clustering accuracy for web-based materials.

Contribution

It introduces a novel approach that integrates named entity information into document vectors for hierarchical clustering, advancing beyond traditional keyword methods.

Findings

Entity-based vectors improve clustering relevance.

Hierarchical clustering using cosine similarity is effective.

Application to web learning materials demonstrates practical utility.

Abstract

Keyword-based information processing has limitations due to simple treatment of words. In this paper, we introduce named entities as objectives into document clustering, which are the key elements defining document semantics and in many cases are of user concerns. First, the traditional keyword-based vector space model is adapted with vectors defined over spaces of entity names, types, name-type pairs, and identifiers, instead of keywords. Then, hierarchical document clustering can be performed using the similarity measure defined as the cosines of the vectors representing documents. Experimental results are presented and discussed. Clustering documents by information of named entities could be useful for managing web-based learning materials with respect to related objects.