Sheldon Klein
Sheldon Klein Professor of Computer Sciences and LinguisticsPh.D., University of California, Berkeley, 1963
Computer Sciences Department
University of Wisconsin
1210 W. Dayton St.
Madison, WI 53706-1685
telephone: (608) 262-1204
fax: (608) 262-9777
email: sklein@cs.wisc.edu
http://www.cs.wisc.edu/~sklein/sklein.html
My current interests are in three related research efforts:
1. The first involves a meta-linguistic, natural language processing system that can be configured to model a variety of theoretical linguistic models. Its semantic structures are in the form of a relational calculus that can be expressed in implicit semantic networks. The basic semantic units are objects and relations. Objects may be atoms of classes, or contain relational structures. Relations may be logical operators. All units in the system are associated with Boolean feature vectors. During the course of generation or recognition, inheritance of features is bi-directional. Because relations can be defined as logical operators, script-like, world knowledge rules can be encoded in the same notation used to map semantic structures to syntactic units. Semantic/syntactic production rules are represented as data, and the same rules can be used for both generation and recognition. The system can contain more than one grammar, thereby allowing it to be configured either as a machine translation system, or as a natural language interface to application command languages.
2. Applications of the above project to the creation and interpretation of 3-D Virtual Reality simulations.
3. The combinatoric problems associated with unrestricted models of human language processing suggest that real-world knowledge systems may have evolved in forms that make combinatoric processing problems linear. Boolean groups and analogy appear to play implortant roles in complex behavioral systems, including natural language. Categorial grammar models are especially amenable to such analyses, and can readily be implemented in connectionist models. This approach may yield a transparent means of linking language structure to neural net theory. The work has implications for the evolutionary history of human cognition, and my research effort has sometimes involved analysis of archaeological materials dating to as early as the Middle-to-Upper Paleolithic transition.
The invention of computationally plausible knowledge systems in the upper Paleolithic, in The Origins of Human Behaviour, R. Foley, ed., pp. 67-81, Unwin Hyman, London, 1991.
Grammars, the I Ching and Levi-Strauss: More on Siemens' `Three Formal Theories of Cultural Analogy', J. Quantitative Anthropology, vol. 6, pp. 263-271, 1996.