Composing Information, Extraction, Semantic Parsing and Tractable Inference for Deep NLP

We developed new information extraction technologies. Our Vinculum entity linker is simple and modular; we compare it to other top systems analyze approaches to mention extraction, candidate generation, entity type prediction, entity coreference, and coherence. We also developed both unsupervised and semi-supervised algorithms for event extraction that exploit parallel news streams, showing significant performance improvements on multiple event extractors over ACE2005 and TAC-KBP 2015 datasets. Finally, we developed new natural language processing tools (eg, semantic parsing) and introduced efficient inference algorithms for extracted knowledge bases.Descriptors: Natural language processing SOFTWARE, SEMATICS, ALGORITHMsSubject Categories: Information ScienceLinguisticsOperations ResearchDistribution Statement: APPROVED FOR PUBLIC RELEASEDEFENSE TECHNICAL INFORMATION CENTER8725 John J. Kingman Road, Fort Belvoir, VA 22060-62181-800-CAL-DTIC (1-800-225-3842)