Moving toward semantics for language processing: recent advances in resource construction and application

Summary form only given. In this paper the author describes the recent work at various locations, focusing on the resource components required (including ISI's symbol definition ontology Omega and the BBN-UPenn-Colorado-ISI large corpus OntoBank of (shallow) meaning representations) and the resources and methods one needs to build them (including existing ontologies, human annotation procedures, and a verification methodology, as embodied in a pilot project called Learning by Reading that involves various prominent researchers). This work is bringing together the NLP and KE/KR communities after a separation of some 40 years, and poses challenges for each side: dealing with large amounts of possibly incorrect knowledge is not something KE/KR has comfortable with, and dealing with requirements that knowledge be formalized and consistent is not something that large-scale NLP has been comfortable with. The research discussed here enables numerous experiments on the wide-scale use of (shallow) semantics, to the benefit of NLP applications of all kinds.