Collective Named Entity Disambiguation using Graph Ranking and Clique Partitioning Approaches

Proceedings of COLING 2014, the 25th International Conference on Computational Linguistics: Technical Papers, pages 1544–1555, Dublin, Ireland, August 23-29 2014. Collective Named Entity Disambiguation using Graph Ranking and Clique Partitioning Approaches Ayman Alhelbawy∗† and Robert Gaizauskas ∗ ∗The University of Sheffield, Regent Court, 211 Portobello Street, Sheffield, S1 4DP, U.K †Faculty of Computers and Information, Fayoum University, Fayoum, Egypt ayman,R.Gaizauskas@dcs.shef.ac.uk Abstract Disambiguating named entities (NE) in running text to their correct interpretations in a specific knowledge base (KB) is an important problem in NLP. This paper presents two collective disam- biguation approaches using a graph representation where possible KB candidates for NE textual mentions are represented as nodes and the coherence relations between different NE candidates are represented by edges. Each node has a local confidence score and each edge has a weight. The first approach uses Page-Rank (PR) to rank all nodes and selects a candidate based on PR score combined with local confidence score. The second approach uses an adapted Clique Par- titioning technique to find the most weighted clique and expands this clique until all NE textual mentions are disambiguated. Experiments on 27,819 NE textual mentions show the effectiveness of both approaches, outperforming both baseline and state-of-the-art approaches. 1 Introduction Named entities (NEs) have received a lot of attention from the NLP community over the last two decades (see, e.g. Nadeau and Sekine (2007)). Most of this work has focussed on the task of recognizing the boundaries of NE mentions in text and classifying them into one of several classes, such as Person, Organization or Location. However, references to entities in the real world are often ambiguous: there is a many-to-many relation between NE mentions in text and the entities denoted by these mentions in the world. For example, the same NE mention “Norfolk” may refer to a person, “Peter Norfolk, a wheelchair tennis player”, a place in the United Kingdom, “Norfolk County”, or a place in the United States like “Norfolk, Massachusetts”; conversely, one entity many be known by many names, such as “Cat Stevens”, “Yusuf Islam” and “Steven Georgiou”. The task of named entity disambiguation (NED) is to establish a correct mapping between each NE mention in a document and the entity it denotes in the real world. Following most researchers in this area, we treat entries in a large knowledge base (KB) as surrogates for real world entities when carrying out NED and, in particular, use Wikipedia as the reference KB against which to disambiguate NE mentions. NED is important for tasks like KB population, where we want to extract new information from text about an entity and add it to a pre-existing entry for that entity in a KB, or for information retrieval where we may want to cluster or filter results for different entities with the same textual mentions. The main hypothesis underlying this work is that different NEs in a document help to disambiguate each other. However, other textual mentions in the document are also ambiguous. So, what is needed is …

Content truncated to 500 words