Inducing Gazetteers for Named Entity Recognition by Large-Scale Clustering of Dependency Relations

We propose using large-scale clustering of dependency relations between verbs and multiword nouns (MNs) to construct a gazetteer for named entity recognition (NER). Since dependency relations capture the semantics of MNs well, the MN clusters constructed by using dependency relations should serve as a good gazetteer. However, the high level of computational cost has prevented the use of clustering for constructing gazetteers. We parallelized a clustering algorithm based on expectationmaximization (EM) and thus enabled the construction of large-scale MN clusters. We demonstrated with the IREX dataset for the Japanese NER that using the constructed clusters as a gazetteer (cluster gazetteer) is a effective way of improving the accuracy of NER. Moreover, we demonstrate that the combination of the cluster gazetteer and a gazetteer extracted from Wikipedia, which is also useful for NER, can further improve the accuracy in several cases.