A Broad-Coverage Deep Semantic Lexicon for Verbs

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 3243–3251 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 3243 A Broad-Coverage Deep Semantic Lexicon for Verbs James Allen1,2, Hannah An1, Ritwik Bose1, Will de Beaumont2 & Choh Man Teng2 1University of Rochester, 2Institute of Human and Machine Cognition 1Dept. Computer Science, Rochester, NY 14627 USA, 240 S. Alcaniz, Pensacola, FL 32501 USA {jallen,wbeaumont,cmteng}@ihmc.us, {yan2,rbose}@cs.rochester.edu Abstract Progress on deep language understanding is inhibited by the lack of a broad coverage lexicon that connects linguistic behavior to ontological concepts and axioms. We have developed COLLIE-V, a deep lexical resource for verbs, with the coverage of WordNet and syntactic and semantic details that meet or exceed existing resources. Bootstrapping from a hand-built lexicon and ontology, new ontological concepts and lexical entries, together with semantic role preferences and entailment axioms, are automatically derived by combining multiple constraints from parsing dictionary definitions and examples. We evaluated the accuracy of the technique along a number of different dimensions and were able to obtain high accuracy in deriving new concepts and lexical entries. COLLIE-V is publicly available. Keywords: semantic lexicon, ontology 1. Introduction While there are significant ongoing efforts to develop deeper language understanding systems, progress is hampered by the lack of a broad-coverage deep semantic lexicon linked to ontologies that can support reasoning about commonsense knowledge. By broad-coverage we mean that the lexicon substantially covers typical English usage (on the scale of WordNet (Fellbaum, 1998)). By deep we mean that all words are assigned senses that are organized into an ontology, and that each sense has associated semantic roles with semantic preferences and syntactic linking templates. By ontology we mean not only a hierarchy of concepts with inheritance of properties, but also axioms that capture the relationships between concepts, especially temporal and causal relationships for events. We have developed Comprehensive OntoLogy and Lexicon In English (COLLIE), a new resource constructed by bootstrapping from an initial hand-built lexicon and ontology, and using a semantic parser to read definitions automatically from dictionaries to extend the coverage. Here we report on the verb component (COLLIE-V) which is the first part of the overall effort to build an extensive semantic lexicon. The full COLLIE, including nominal concepts and properties, will be released at a later date. As an example showing the typical contents of COLLIE-V, consider the entry for one sense of the verb kill, which we will call ONT::KILL. The information about this sense is captured in two places: (1) information in the ontology about the concept ONT::KILL (which contains other verbs such as murder and slaughter in addition to kill), and (2) information in the lexicon about the behavior of the verb kill. Looking first at the concept, ONT::KILL is a subconcept of ONT::DESTROY, which involves events of destruction. ONT::KILL has two core semantic roles, AGENT (playing a causal role in the event) and AFFECTED (the object affected by the event). Furthermore, the roles are associated with semantic preferences. …

Content truncated to 500 words