Automatic Classification of Geographical Named Entities

Performing accurate Named Entity (NE) classification (NEC) has recently become a central issue in many NLP applications, such as Information Extraction and Question Answering, among others. Most state-of-the-art NEC systems use coarse-grained MUC-style datasets for performing the NEC task reducing it to distinguish among LOCATION, PERSON, ORGANIZATION and so. There is, however, a growing interest on using finer-grained classification sets. This paper describes a methodology that applies Machine Learning techniques for a finer-grained classification of NEs that have been previously classified as locations by a NERC system.