BaseNPs that contain gene names: domain specificity and genericity

The names of named entities very often occur as constituents of larger noun phrases which denote different types of entity. Understanding the structure of the embedding phrase can be an enormously beneficial first step to enhancing whatever processing is intended to follow the named entity recognition in the first place. In this paper, we examine the integration of general purpose linguistic processors together with domain specific named entity recognition in order to carry out the task of baseNP detection. We report a best F-score of 87.17% on this task. We also report an inter-annotator agreement score of 98.8 Kappa on the task of baseNP annotation of a new data set.