skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Detecting knowledge base inconsistencies using automated generation of text and examples

Abstract

Verifying the fidelity of domain representation in large knowledge bases (KBs) is a difficult problem: domain experts are typically not experts in knowledge representation languages, and as knowledge bases grow more complex, visual inspection of the various terms and their abstract definitions, their interrelationships and the limiting, boundary cases becomes much harder. This paper presents an approach to help verify and refine abstract term definitions in knowledge bases. It assumes that it is easier for a domain expert to determine the correctness of individual concrete examples than it is to verify and correct all the ramifications of an abstract, intentional specification. To this end, our approach presents the user with an interface in which abstract terms in the KB are described using examples and natural language generated from the underlying domain representation. Problems in the KB are therefore manifested as problems in the generated description. The user can then highlight specific examples or parts of the explanation that seem problematic. The system reasons about the underlying domain model by using the discourse plan generated for the description. This paper briefly describes the working of the system and illustrates three possible types of problem manifestations using an example of a specificationmore » of floating-point numbers in Lisp.« less

Authors:
;  [1]
  1. Univ. of Pittsburgh, PA (United States)
Publication Date:
OSTI Identifier:
430698
Report Number(s):
CONF-960876-
TRN: 96:006521-0073
Resource Type:
Conference
Resource Relation:
Conference: 13. National conference on artifical intelligence and the 8. Innovative applications of artificial intelligence conference, Portland, OR (United States), 4-8 Aug 1996; Other Information: PBD: 1996; Related Information: Is Part Of Proceedings of the thirteenth national conference on artificial intelligence and the eighth innovative applications of artificial intelligence conference. Volume 1 and 2; PB: 1626 p.
Country of Publication:
United States
Language:
English
Subject:
99 MATHEMATICS, COMPUTERS, INFORMATION SCIENCE, MANAGEMENT, LAW, MISCELLANEOUS; ARTIFICIAL INTELLIGENCE; KNOWLEDGE BASE; ERRORS; LEARNING; NATURAL LANGUAGE

Citation Formats

Mittal, V.O., and Moore, J.D. Detecting knowledge base inconsistencies using automated generation of text and examples. United States: N. p., 1996. Web.
Mittal, V.O., & Moore, J.D. Detecting knowledge base inconsistencies using automated generation of text and examples. United States.
Mittal, V.O., and Moore, J.D. 1996. "Detecting knowledge base inconsistencies using automated generation of text and examples". United States. doi:.
@article{osti_430698,
title = {Detecting knowledge base inconsistencies using automated generation of text and examples},
author = {Mittal, V.O. and Moore, J.D.},
abstractNote = {Verifying the fidelity of domain representation in large knowledge bases (KBs) is a difficult problem: domain experts are typically not experts in knowledge representation languages, and as knowledge bases grow more complex, visual inspection of the various terms and their abstract definitions, their interrelationships and the limiting, boundary cases becomes much harder. This paper presents an approach to help verify and refine abstract term definitions in knowledge bases. It assumes that it is easier for a domain expert to determine the correctness of individual concrete examples than it is to verify and correct all the ramifications of an abstract, intentional specification. To this end, our approach presents the user with an interface in which abstract terms in the KB are described using examples and natural language generated from the underlying domain representation. Problems in the KB are therefore manifested as problems in the generated description. The user can then highlight specific examples or parts of the explanation that seem problematic. The system reasons about the underlying domain model by using the discourse plan generated for the description. This paper briefly describes the working of the system and illustrates three possible types of problem manifestations using an example of a specification of floating-point numbers in Lisp.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = 1996,
month =
}

Conference:
Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.

Save / Share:
  • In this paper, we present a conceptual analysis of knowledge-base development methodologies. The purpose of this research is to help overcome the high cost and lack of efficiency in developing knowledge base representations for artificial intelligence applications. To accomplish this purpose, we analyzed the available methodologies and developed a knowledge-base development methodology taxonomy. We review manual, machine-aided, and machine-learning methodologies. A set of developed characteristics allows description and comparison among the methodologies. We present the results of this conceptual analysis of methodologies and recommendations for development of more efficient and effective tools.
  • In this paper, we present a conceptual analysis of knowledge-base development methodologies. The purpose of this research is to help overcome the high cost and lack of efficiency in developing knowledge base representations for artificial intelligence applications. To accomplish this purpose, we analyzed the available methodologies and developed a knowledge-base development methodology taxonomy. We review manual, machine-aided, and machine-learning methodologies. A set of developed characteristics allows description and comparison among the methodologies. We present the results of this conceptual analysis of methodologies and recommendations for development of more efficient and effective tools.
  • An Automated MEhsing Knowledge System that automatically generates quandrilateral finite element meshes is discussed in this paper. This system, AMEKS, operates in conjunction with existing algorithmic software on a large class of two-dimensional bodies given only their boundaries as input. These bodies can consist of multiple materials and can have both curved and straight boundaries. Prior to AMEKS, this task required that a skilled analyst spend a few hours to a few days decomposing each complex body into primitive regions and preparing each primitive for algorithmic meshing, based on problem geometry and mesh requirements. With AMEKS, quadrilateral meshes comparable tomore » those of experienced analysts are generated automatically in a small fraction of the time formerly required.« less
  • Intelligent aids to the development of scientific software are in great demand but in short supply. Two key components of such expert systems will be automated reasoning and knowledge representation methodologies. First order logic is described as the common language of a class of reasoning systems and database design methods. An overview is given of reasoning paradigms used in current expert systems and of database design methods appropriate to creation of the knowledge base to support such reasoning. Finally, some of the problems seen as topics for further research in this area are described. 14 refs.
  • This paper presents a knowledge system approach to automated 2-D quadrilateral mesh generation. This approach mimics the most popular technique currently used by analysts for all-quadrilateral mesh generation, that of decomposing an arbitrary geometry into subregions for mapping. We enumerate the merits of this approach over other automated mesh generation techniques and discuss the basic philosophy and structure of the knowledge system and its tie to enhanced mapping algorithms. Examples of automatically generated meshes are presented and compared with analysts' generated meshes as an indication of the capability of this approach. 15 refs. 12 figs.