tutorial

OntoNotes: The 90% SolutionSameer S. Pradhan and Nianwen XueOntoNotes is a five year multi-site collaboration between BBN Technologies, Information Sciences In-stitute of University of Southern California, University of Colorado, University of Pennsylvania and Bran-deis University. The goal of the OntoNotes project is to provide linguistic data annotated with a skeletalrepresentation of the literal meaning of sentences including syntactic parse, predicate-argument structure,coreference, and word senses linked to an ontology, allowing a new generation of language understandingtechnologies to be developed with new functional capabilities.In its third year of existence, the OntoNotes project has generated a large amount of high quality datacovering various layers of linguistic annotation. This is probably the first time that data of such quality hasbeen available in large quantities covering multiple genres (newswire, broadcast news, broadcast conver-sation and weblogs) and languages (English, Chinese and Arabic). The guiding principle has been to finda “sweet spot” in the space of inter-tagger agreement, productivity, and depth of representation. The mosteffectiveuseofthisresourceforresearchrequiressimultaneousaccesstomultiplelayersofannotation. Thishas been made possible by representing the corpus with a relational database to accommodate the denseconnectedness of the data and ensure consistency across layers. In order to facilitate ease of understandingand ...