Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Incremental interpretation
Pereira F., Pollack M. Artificial Intelligence50 (1):37-82,1991.Type:Article
Date Reviewed: Aug 1 1992

The semantic interpretation of utterances in a discourse is discussed. The paper falls within computational linguistics, but has a strong AI vein. As the authors rightly point out, compositionality and incrementality are desiderata for any formalism for semantic interpretation. The authors show that the traditional approach to semantic interpretation, Montague semantics, achieves them only by sacrificing simplicity of description. DRT, on the other hand, is not incremental.

Their own formalism rests on the idea of integrating discourse and domain knowledge into the interpretation process. The basic structures are so-called conditional expressions, consisting of a sense and a set of assumptions. Rules are used to manipulate these structures. The formalism is implemented in a system for interactive knowledge acquisition, called CANDIDE. The paper starts by giving a short introduction to the formalism and its processing. This introduction includes some general remarks about the interaction with the discourse model (designed after Barbara Grosz). Then some modestly complex examples demonstrate its use. The authors conclude with a short comparison to other  approaches. 

Unfortunately, the authors--as they are aware--propose only a simple processing model. A few hints are added where this model will not suffice. Since rule application is non-monotonic, this shortcoming is important. Also, the treatment of complex determiners is mentioned only in passing. Clearly, much more work is needed before the appropriateness of the formalism can be judged.

Semantic interpretation is a central point of interest for natural language processing. So far, traditional linguistic approaches have not been very successful in realistic applications. Pure AI approaches, on the other hand, tend to lack generality and a sound theoretical basis. The approach of the authors, which tries to combine the strong points of both approaches, looks promising enough that I recommend that anyone working in this field study it.

Reviewer:  H. Trost Review #: CR115754
Bookmark and Share
 
Natural Language Processing (I.2.7 )
 
 
Representations (Procedural And Rule-Based) (I.2.4 ... )
 
Would you recommend this review?
yes
no
Other reviews under "Natural Language Processing": Date
Current research in natural language generation
Dale R. (ed), Mellish C. (ed), Zock M., Academic Press Prof., Inc., San Diego, CA, 1990. Type: Book (9780122007354)
Nov 1 1992
Natural language and computational linguistics
Beardon C., Lumsden D., Holmes G., Ellis Horwood, Upper Saddle River, NJ, 1991. Type: Book (9780136128137)
Jul 1 1992
Cooperative interfaces to information systems
Bolc L. (ed), Jarke M., Springer-Verlag New York, Inc., New York, NY, 1986. Type: Book (9780387165998)
Apr 1 1988
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy