A Preliminary Evaluation of Word Representations for Named-Entity Recognition

We use different word representations as word features for a named-entity recognition (NER) system with a linear model. This work is part of a larger empirical survey, evaluating different word representations on different NLP tasks. We evaluate Brown clusters, Collobert and Weston (2008) embeddings, and HLBL (Mnih & Hinton, 2009) embeddings of words. All three representations improve accuracy on NER, with the Brown clusters providing a larger improvement than the two embeddings, and the HLBL embeddings more than the Collobert and Weston (2008) embeddings. We also discuss some of the practical issues in using embeddings as features. Brown clusters are simpler than embeddings because they require less hyperparameter tuning.