Graph Based Disambiguation of Named Entities using Linked Data

Abstract

Identifying entities such as people, organizations, songs, or places in natural language texts is needful for semantic search, machine translation, and information extraction. A key challenge is the ambiguity of entity names, requiring robust methods to disambiguate names to the entities registered in a knowledge base. Several approaches aim to tackle this problem, they still achieve poor accuracy. We address this drawback by presenting a novel knowledge-base-agnostic approach for named entity disambiguation. Our approach includes the HITS algorithm combined with label expansion strategies and string similarity measure like the n-gram similarity. Based on this combination, we can efficiently detect the correct URIs for a given set of named entities within an input text.