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

Title: Scaling up explanation generation: Large-scale knowledge bases and empirical studies

Abstract

To explain complex phenomena, an explanation system must be able to select information from a formal representation of domain knowledge, organize the selected information into multisentential discourse plans, and realize the discourse plans in text. Although recent years have witnessed significant progress in the development of sophisticated computational mechanisms for explanation, empirical results have been limited. This paper reports on a seven year effort to empirically study explanation generation from semantically rich, large-scale knowledge bases. We first describe Knight, a robust explanation system that constructs multi-sentential and multi-paragraph explanations from the Biology Knowledge Base, a large-scale knowledge base in the domain of botanical anatomy, physiology, and development. We then introduce the Two Panel evaluation methodology and describe how Knight`s performance was assessed with this methodology in the most extensive empirical evaluation conducted on an explanation system. In this evaluation, Knight scored within {open_quotes}half a grade{close_quote} of domain experts, and its performance exceeded that of one of the domain experts.

Authors:
 [1];  [2]
  1. North Carolina State Univ., Raleigh, NC (United States)
  2. Univ. of Texas, Austin, TX (United States)
Publication Date:
OSTI Identifier:
430689
Report Number(s):
CONF-960876-
TRN: 96:006521-0064
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; COMPUTER-AIDED INSTRUCTION; KNOWLEDGE BASE; ARTIFICIAL INTELLIGENCE

Citation Formats

Lester, J.C., and Porter, B.W.. Scaling up explanation generation: Large-scale knowledge bases and empirical studies. United States: N. p., 1996. Web.
Lester, J.C., & Porter, B.W.. Scaling up explanation generation: Large-scale knowledge bases and empirical studies. United States.
Lester, J.C., and Porter, B.W.. 1996. "Scaling up explanation generation: Large-scale knowledge bases and empirical studies". United States. doi:.
@article{osti_430689,
title = {Scaling up explanation generation: Large-scale knowledge bases and empirical studies},
author = {Lester, J.C. and Porter, B.W.},
abstractNote = {To explain complex phenomena, an explanation system must be able to select information from a formal representation of domain knowledge, organize the selected information into multisentential discourse plans, and realize the discourse plans in text. Although recent years have witnessed significant progress in the development of sophisticated computational mechanisms for explanation, empirical results have been limited. This paper reports on a seven year effort to empirically study explanation generation from semantically rich, large-scale knowledge bases. We first describe Knight, a robust explanation system that constructs multi-sentential and multi-paragraph explanations from the Biology Knowledge Base, a large-scale knowledge base in the domain of botanical anatomy, physiology, and development. We then introduce the Two Panel evaluation methodology and describe how Knight`s performance was assessed with this methodology in the most extensive empirical evaluation conducted on an explanation system. In this evaluation, Knight scored within {open_quotes}half a grade{close_quote} of domain experts, and its performance exceeded that of one of the domain experts.},
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:
  • Knowledge workers need tools to help them navigate through, evaluate, and understand large stores of information. Motivated by the needs of ARDA's, Novel Intelligence from Massive Data program, Battelle, Stanford University, and IBM have developed a suite of technologies for knowledge discovery, knowledge extraction, knowledge representation, automated reasoning, explanation, and human information interaction. Our team has developed an integrated analytic environment composed of a collection of analyst associates, software components that aid the analyst at different stages of the analytical process, collectively known as 'Knowledge Associates for Novel Intelligence (KANI)'. As part of this effort, we have incorporated a Querymore » Answering and Explanation component that allows analysts to pose questions of the system based on the knowledge it has of a particular domain and specific tasking (problem). Answers are presented along with optional information about sources, assumptions, explanation summaries, and interactive justifications. This paper describes the analyst requirements and response to the explanation component of the KANI system. We believe the explanation infrastructure, its interface for analysts and knowledge workers, and the provenance requirements are all contributions that are leverageable beyond the KANI implementation.« less
  • At this time, a number of high-qualified expert teams in different countries have accumulated significant experience in working up and introducing proficient algorithms of automated control, diagnostics, monitoring, on-line informative support, and off-line analysis for large steam turbines of fossil-fired and nuclear power plant units at their transients. In particular, such works were carried out at All-Russia Thermal Engineering Research Institute in Moscow. When put into effect, these developments do improve plant unit operation and maintenance. The creation of these algorithms is based on deep technological research of turbine transients with the use of their mathematical modeling. Currently, the centralmore » problem becomes a reiteration and adaptation of these developments to other objects which differ in design, scheme, and operation features. This problem is to be solved at the least cost. For certain complex tasks, it seems to be expedient to use expert system methodology with special knowledge bases. Along with the selection of such tasks, the creation of more or less general knowledge bases requires us to stratify the potential objects according to their principle features. The proposed forms of a knowledge presentation, including the mathematical models and logical rules, should correspond to the considered tasks and objects, as well as to the program tools to be applied. Such developments have been carried out for some tasks of the operating informative support and post-operating analysis of large steam turbine transients as well as their mathematical modeling.« less
  • Abstract not provided.
  • The UCC-ND Cost Estimating System requires the storage, maintenance, and retrieval of many data bases varying widely in size (up to 50,000 records) and associated with each other in a variety of subtly implied ways. A system was devised whereby the data are stored on a Librarian master file in a ''highly atomized'' form and are retrieved and configured at job submission time on the basis of criteria such as work-package structuring, Engineering Department organizational lines, etc. The simplicity and ease-of-use of Librarian enabled implementation of an on-line data base management system in a period of approximately two months, andmore » led to higher benchmarks of responsiveness and reliability than were previously possible. 5 figures.« less
  • Environmental monitoring and restoration at the Department of Energy`s Savannah River Site (SRS) requires efficient access to large amounts of diverse spatial data. In order to handle this task the Environmental Sciences Section (ESS) of the Westinghouse Savannah River Company (WSRC) created an environmental data atlas that would use spatial keys to link all data sources to a common data base. These data include extensive information relating to both environmental and man-made features. The data were available in a multitude of different data structures, coordinate systems and formats. The final database also included multimedia information such as airborne multispectral scannermore » data, aircraft video, photography, bibliographic data, and scanned map bases. It was important that all of the data be readily accessible on the desktop of scientists regardless of the type of computer system they used. The purpose of this paper is to describe the creation of this extensive digital environmental data atlas.« less