Fine-Grained Entity Recognition

Entity Recognition (ER) is a key component of relation extraction systems and many other natural-language processing applications. Unfortunately, most ER are restricted to produce labels from small set entity classes, e.g., person, organization, location or miscellaneous. In order intelligently understand text extract wide range information, it useful more precisely determine the semantic classes entities mentioned in unstructured text. This paper defines fine-grained 112 tags, formulates tagging problem as multi-class, multi-label classification, describes an unsupervised method for collecting training data, presents FIGER implementation. Experiments show that system accurately predicts tags entities. Moreover, provides information system, increasing F1 score by 93%. We make its data available resource future work.