An Approach to Named Entity Disambiguation Based on Explicit Semantics

Identification of the named entities in an unstructured, human written text is a well established subtask of Natural Language Processing. However, marking each occurrence of a named entity in the text with a class label is usually not sufficient, as the same word often describe several different entities. A dedicated area of NLP, Named Entity Disambiguation, has been devised to solve this problem. In our work we present an approach to the problem of Named Entity Disambiguation based on the Explicit Semantic Analysis. We use a semantic similarity measure based on the similarity between context of the entity and the documents describing the possible meanings. We use an additional semantics provided by Wikipedia, such as disambiguation and redirect pages or links between the documents. Evaluation of the proposed method shows an improvement over the traditionally used Latent Semantic Analysis.