Proceedings of the Second ACL Workshop on Effective Tools and Methodologies for Teaching NLP and CL}
@Book{TNLP:2005, editor = {Chris Brew and Dragomir Radev}, title = {Proceedings of the Second ACL... more @Book{TNLP:2005, editor = {Chris Brew and Dragomir Radev}, title = {Proceedings of the Second ACL Workshop on Effective Tools and Methodologies for Teaching NLP and CL}, month = {June}, year = {2005}, address = {Ann Arbor, Michigan}, publisher = {Association for Computational Linguistics}, url = {http://www.aclweb.org/anthology/W/W05/W05-01} } @InProceedings{hearst: 2005:TNLP, author = {Hearst, Marti}, title = {Teaching Applied Natural Language Processing: Triumphs and Tribulations}, booktitle = {Proceedings of the Second ACL ...
This paper introduces a new model of selectional preference induction.Unlike previous approaches ... more This paper introduces a new model of selectional preference induction.Unlike previous approaches we provide a stochastic generationmodel for the words that appear as arguments of a predicate. Morespecifically, we define a hidden Markov model with the general shapeof our semantic class hierarchy and we use an EM-like algorithm forparameter estimation. This model enables us to handle word senseambiguity in the
... David R. Traum 1 Lenhart K. Schubert 2 Massimo Poesio 3 Nathaniel G. Martin 2 Marc Light 4 Ch... more ... David R. Traum 1 Lenhart K. Schubert 2 Massimo Poesio 3 Nathaniel G. Martin 2 Marc Light 4 Chung Hee Hwang 2 Peter Heeman 2 George Ferguson 2 James F. Allen 2 Abstract We describe the goals, architecture, and functioning of the ...
Robustness is a highly desirable trait for natural language processing systems. Among other thing... more Robustness is a highly desirable trait for natural language processing systems. Among other things, a robust system would have broad coverage and its output would degrade gracefully as inputs strayed from the area of coverage. Systems that attempt global opti- mization and complete syntactic and semantic analysis tend to have limited coverage and to degrade less than gracefully. In addition, as their coverage increases, their efficiency often decreases. In this paper we describe the CHUMP system which utilizes partial parsing and underspecified semantic representations to provide partial semantic analyses for spoken utterances robustly and efficiently. These two characteristics are very relevant for corpus-oriented semantic anal- ysis and thus although the CHUMP system was built to be part of a speech-to-speech translation system, its architecture could be used for corpus-oriented semantic analysis.
Increasingly, inheritance hierarchies are being used to reduce redundancy in natural language pro... more Increasingly, inheritance hierarchies are being used to reduce redundancy in natural language processing lexicons. Systems that utilize inheritance hierarchies need to be able to insert words under the optimal set of classes in these hierarchies. In this paper, we formalize this problem for feature-based default inheritance hierarchies. Since the problem turns out to be NP-complete, we present an approximation algorithm for it. We show that this algorithm is efficient and that it performs well with respect to a number of standard problems for default inheritance. A prototype implementation has been tested on lexical hierarchies and it has produced encouraging results. The work presented here is also relevant to other types of default hierarchies.
Proceedings of the Second ACL Workshop on Effective Tools and Methodologies for Teaching NLP and CL}
@Book{TNLP:2005, editor = {Chris Brew and Dragomir Radev}, title = {Proceedings of the Second ACL... more @Book{TNLP:2005, editor = {Chris Brew and Dragomir Radev}, title = {Proceedings of the Second ACL Workshop on Effective Tools and Methodologies for Teaching NLP and CL}, month = {June}, year = {2005}, address = {Ann Arbor, Michigan}, publisher = {Association for Computational Linguistics}, url = {http://www.aclweb.org/anthology/W/W05/W05-01} } @InProceedings{hearst: 2005:TNLP, author = {Hearst, Marti}, title = {Teaching Applied Natural Language Processing: Triumphs and Tribulations}, booktitle = {Proceedings of the Second ACL ...
This paper introduces a new model of selectional preference induction.Unlike previous approaches ... more This paper introduces a new model of selectional preference induction.Unlike previous approaches we provide a stochastic generationmodel for the words that appear as arguments of a predicate. Morespecifically, we define a hidden Markov model with the general shapeof our semantic class hierarchy and we use an EM-like algorithm forparameter estimation. This model enables us to handle word senseambiguity in the
... David R. Traum 1 Lenhart K. Schubert 2 Massimo Poesio 3 Nathaniel G. Martin 2 Marc Light 4 Ch... more ... David R. Traum 1 Lenhart K. Schubert 2 Massimo Poesio 3 Nathaniel G. Martin 2 Marc Light 4 Chung Hee Hwang 2 Peter Heeman 2 George Ferguson 2 James F. Allen 2 Abstract We describe the goals, architecture, and functioning of the ...
Robustness is a highly desirable trait for natural language processing systems. Among other thing... more Robustness is a highly desirable trait for natural language processing systems. Among other things, a robust system would have broad coverage and its output would degrade gracefully as inputs strayed from the area of coverage. Systems that attempt global opti- mization and complete syntactic and semantic analysis tend to have limited coverage and to degrade less than gracefully. In addition, as their coverage increases, their efficiency often decreases. In this paper we describe the CHUMP system which utilizes partial parsing and underspecified semantic representations to provide partial semantic analyses for spoken utterances robustly and efficiently. These two characteristics are very relevant for corpus-oriented semantic anal- ysis and thus although the CHUMP system was built to be part of a speech-to-speech translation system, its architecture could be used for corpus-oriented semantic analysis.
Increasingly, inheritance hierarchies are being used to reduce redundancy in natural language pro... more Increasingly, inheritance hierarchies are being used to reduce redundancy in natural language processing lexicons. Systems that utilize inheritance hierarchies need to be able to insert words under the optimal set of classes in these hierarchies. In this paper, we formalize this problem for feature-based default inheritance hierarchies. Since the problem turns out to be NP-complete, we present an approximation algorithm for it. We show that this algorithm is efficient and that it performs well with respect to a number of standard problems for default inheritance. A prototype implementation has been tested on lexical hierarchies and it has produced encouraging results. The work presented here is also relevant to other types of default hierarchies.
Uploads
Papers by Marc Light