Working with Visionaries on the Frontlines of Scientific Progress Worldwide
Nuance Foundation Grants

Automatic Acquisition of a Deep Semantic Lexicon

University of Rochester - Department of Computer Science

University of Rochester - Department of Computer Science

The University of Rochester is one of the country's top-tier research universities. Its 158 buildings house more than 200 academic majors, more than 2,000 faculty and instructional staff, and some 10,500 students—approximately half of whom are women.

The University of Rochester Department of Computer Science is home to cutting-edge research in artificial intelligence, computer systems, human-computer interaction, and theory of computation. It offers a PhD program, an MS degree, and BS and BA undergraduate majors.

Project: Automatic Acquisition of a Deep Semantic Lexicon

James Allen, Jansen Orfan, Omid Bakhshandeh

While researchers agree that deep understanding will require combining natural language processing with knowledge and reasoning systems, very little effort is currently being made to bring these areas together in a serious way. One main reason for this is a lack of extensive semantic lexical resources that connect words to knowledge bases. The project will develop novel methods for automatically constructing a broad coverage deep semantic lexicon for English simultaneously with constructing a commonsense knowledge base. The lexicon will be organized by Wordnet senses, and associated with a commonsense knowledge base expressed in a well recognized knowledge representation standard, namely OWL-DL. To accomplish this ambitious goal, we are leveraging a unique resource, namely the TRIPS language understanding system. TRIPS provides broad-coverage parsing of English into a rich logical form constructed from a hand-built ontology, lexicon and grammar. TRIPS is the result of several decades of research and experimentation. While the core lexicon is only about 10,000 words, the ontology is fully connected into WordNet and the system has the ability to build new semantic lexical entries for unknown words on the fly using the WordNet hypernym hierarchy.

Publications:  

Orfan, J. and J. Allen (2013). Toward Learning High-Level Semantic Frames from Definitions. 2nd Annual Conference on Advances in Cognitive Systems, Baltimore, MD.

Orfan, J. and J. Allen (2015) Learning New Relations from Concept Ontologies Derived from Definitions, Twelfth International Symposium on Logical Formalizations of Commonsense Reasoning, to appear.

Bahkshandeh, Omid and J. Allen (2015) From Adjective Glosses to Attribute Concepts: Learning Different Aspects That an Adjective Can Describe. The 11th International Conference on Computational Semantics (IWCS 2015), to appear.

Programs & Grants

100 years ago nobody would have imagined that it may make sense to talk to machines. Today, in the days of speech recognition and speech synthesis to be found in cars, computers, phones and many other devices this is already normal. But it doesn't stop there.

Learn More