Hostname: page-component-5c6d5d7d68-qks25 Total loading time: 0 Render date: 2024-08-06T21:02:06.117Z Has data issue: false hasContentIssue false
  • English
  • Français

Statistical applications of artificial intelligence and knowledge engineering

Published online by Cambridge University Press:  07 July 2009

William A. Gale
William A. Gale
Affiliation:
AT&T Bell Laboratories, 20278, 600 Mountain Avenue, Murray Hill
Get access Rights & Permissions [Opens in a new window]

Abstract

Knowledge engineering (KE) has now provided some effective techniques for formalization of knowledge about goals and actions. These techniques could open new areas of research to statisticians. Experimental systems designed to assist users of statistics have been constructed in experiment design, data analysis, technique application, and technique selection. Knowledge formalization has also been used in experimental programs to assist statisticians in doing data analysis and in building consultation systems. The best-explored application of KE techniques is building consultation systems. It is now a promising area for development. Analogies with successful artificial intelligence AI applications in other fields suggest other statistical applications worth exploring. Opening new areas to research and providing new tools to users would make considerable changes in the use and production of statistical techniques. However, applying currently available KE techniques will lead to more work for statisticians, not less.

Type
Research Article
Information
The Knowledge Engineering Review , Volume 2 , Issue 4 , December 1987 , pp. 227 - 247
Copyright
Copyright © Cambridge University Press 1987

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

Chowdhury, S I, 1987. “State of the art in statistical expert systems” in 7th International Workshop: Expert Systems & Their Applications, Avignon: France. (This paper has half page summaries of 17 previous papers).Google Scholar
Gale, W A, (ed.) 1986a. Artificial Intelligence and Statistics Reading, Massachusetts: Addison Wesley. This is the first book published in the field. It is a logical starting point for learning about the field.Google Scholar
Gale, W A, 1986e. “Overview,” Chapter 1 of Artificial Intelligence and Statistics, Reading, Massachusetts. Addison Wesley. The chapter reviews all papers that I knew about at the time, because they were only available in conference proceedings.Google Scholar
Hahn, G J, 1985. “More intelligent statistical software and statistical expert systems: future directions” The American Statistician 39 pp. 1–16. This was the first major review. It is directed to statisticians, and contains suggestions for research directions.Google Scholar
Hand, D J, 1986b. “Expert systems in statistics” The Knowledge Engineering Review 1 pp. 2–10. Hand's papers have typically a considerable theoretical framework to which their substance is related. This review paper is no exception, doing a nice job of describing aspects of the statistical domain that the KE should appreciate.Google Scholar
Aikins, J S, 1983. “Prototypical knowledge for expert systems,” Artificial Intelligence 120 pp. 163–210. This paper describes Centaur, Aikins' thesis project. It provided a single framework that combined both rules and frames.Google Scholar
de Kleer, J and Brown, J S, 1985. “A Qualitative physics based on confluences,” In: Hobbs and Moore, 1985 pp. 109–194. This paper introduces a calculus of directions, and applies it in a few examples.Google Scholar
Gale, W A, 1986d. “Knowledge-based knowledge acquisition for a statistical consulting system,” International Journal of Man-Machine Studies 26 pp. 55–64. This paper describes the abstract knowledge engineering principles on which student is based.CrossRefGoogle Scholar
Gale, W A and Lubinsky, D, 1986. “A comparison of representations for statistical strategies” In: Proceedings of the American Statistical Association, Statistical Computing Section, Arlington, Virginia: American Statistical Association, pp. 88–96. The paper compares the knowledge representations used by Student and TESS.Google Scholar
Hand, D J, 1986a. “Patterns in statistical strategy” In: Artificial Intelligence and Statistics, Reading, Massachusetts: Addison Wesley pp. 355–388. This paper gives a theoretical view of the task undertaken by the various systems referenced below under strategy of data analysis.Google Scholar
Hayes, P, 1985. “Naive physics I: ontology for liquids” In: Formal Theories of the Commonsense World, Hobbs, and Moore, 1985. Eds, New Jersey: Ablex, Norwood, pp. 71–108. The paper introduces a formalization for the behaviour of liquids.Google Scholar
Hobbs, J, and Moore, R, 1985. Formal Theories of the Commonsense World, New Jersey: Ablex, Norwood. This book is a good introduction to building formal theories in areas not previously formalized.Google Scholar
Newell, A, 1981. “The knowledge level” AI Magazine 2(2) pp. 1–20. His presidential address introduced an idea that has organized further research.Google Scholar
O'Keefe, R, 1985. Logic and Lattices for A Statistical Advisor, Thesis, University of Edinburgh. This thesis describes a system that selects statistical analysis techniques but the greater value is a beginning on a formalization of data description and experiment design.Google Scholar
Oldford, W and Peters, S, 1986a. “Implementation and study of statistical strategy” In: Artificial Intelligence and Statistics, Gale, N A, ed. Reading, Massachusetts: Addison-Wesley, pp. 335–353. This paper gave a theoretical view of strategy for data analysis that is still useful to read.Google Scholar
Oldford, W and Peters, S, 1986c. “Object oriented data representations for statistical data analysis” In COMPSTAT 1986: Proceedings in Computational Statistics, De Antoni, F, Lauro, N and Rizzi, A, (eds.) Vienna: Physica-Verlag pp. 301–308. This paper describes the knowledge representation ideas behind DINDE.Google Scholar
Piaget, J and Inhelder, B, 1951. La genèse de L'idée de hasard chez l'enfant, Presses Universitaires de France: translated by Leake, Burrell, and Fishbein, 1975, The Origin of the Idea of Chance in Children, New York, Norton. A book that describes experiments showing the developmental stages of ideas of randomness and order.Google Scholar
Anderson, J, Farrell, R and Sauers, R, 1984. “Learning to program in LISP,” Cognitive Science 8 pp. 87–129. The paper describes an intelligent tutoring system which taught some basics of programming in LISP.Google Scholar
Barzilay, A, 1984. An Expert System for Tutoring Probability Theory, Ph.D. Thesis submitted to Graduate School of Business, University of Pittsburgh. The system SPIRIT was built using OPS5. It tutors in conditional probabilities.Google Scholar
Blum, R L, 1982. Discovery and Representation of Causal Relationships from a Large Time-Oriented Clinical Database: The RX Project, New York: Springer-Verlag. The book is based on Blum's thesis. It describes the whole RX system.CrossRefGoogle Scholar
Dambroise, E, 1987. Muse: Multivariate Expertise. Thesis INRA, 9 Place Pierre Viala, 34060 Montpelier Cedex, France: an abbreviated report in English is available in: Dambroise E and Massotte, P “Muse, an expert system in statistics” In: COMPSTAT 1986: Proceedings in Computational Statistics, De Antoni, F, Lauro, N and Rizzi, A, (eds.) Vienna: Physica-Verlag, pp. 271–276. The thesis provides Dambroise with enough space to explain MUSE in detail, but the paper is much easier to get.Google Scholar
Haaland, P D, Yen, D and Liddle, R F, 1986. “An expert system for experimental design” In: Proceedings of the Statistical Computing Section, Arlington, Virginia: American Statistical Association, pp. 78–87. The paper describes a system which helps an industrial engineer select an experimental design.Google Scholar
Portier, K M and Lai, P, 1983. “A statistical expert system for analysis determination” Proceedings of the Statistical Computing Section, Arlington, Virginia: American Statistical Association, pp. 309–311. This paper describes the behaviour of a system based on a decision tree.Google Scholar
Stefik, M J, 1980. “Planning with constraints” Report No. 80–784, Computer Science Department, Stanford University, (Doctoral thesis). This thesis shows how knowledge about molecular genetics can be used to constrain plans and reduce an intractable search space.Google Scholar
Sussman, G J, 1975. A Computer Model of Skill Acquisition, New York: Elsevier. This thesis described a program that would learn to write better programs by analyzing bugs in its programs.Google Scholar
Weiner, J M, Horowitz, R and Bauer, M, 1987. “An expert system for analysis of clinical trial data” In: Heiberger, R, (Ed.) Computer Science and Statistics, Proceedings of the Nineteenth Symposium on the Interface, North-Holland. This paper is not so much on an expert system as on the person/machine organization of the large body of knowledge required to design ethical, legal, and knowledge enhancing experiments on treatment of diseases.Google Scholar
Berzuini, C, Ross, G and Larizza, C, 1986. “Developing intelligent software for non-linear model fitting as an expert system” In: COMPSTAT 1986: Proceedings in Computational Statistics, De Antoni, F, Lauro, N and Rizzi, A, (Eds.) Vienna: Physica-Verlag, pp. 259–264. This is a first paper on a project to make an intelligent interface to the “Maximum likelihood program,” a program encoding an advanced statistical procedure.Google Scholar
Carlsen, F and Heuch, I, 1986. “Express—an expert system utilizing standard statistical packages” In: COMPSTAT 1986: Proceedings in Computational Statistics, De Antoni, F, Lauro, N and Rizzi, A, (Eds.) Vienna: Physica-Verlag, pp. 265–270. This system compared the locations of two samples. It used BMDP for calculations, and introduced simulation as a means of testing the system.Google Scholar
Darius, P, 1986. “Building expert systems with the help of existing statistical software” In: COMPSTAT 1986: Proceedings in Computational Statistics, De Antoni, F, Lauro, N and Rizzi, A, (Eds.) Vienna: Physica-Verlag, pp. 277–282. Darius used SAS as the programming language for a rule interpreter, and used that to select a test for comparing two group means.Google Scholar
Drapier, P, 1987. “Le système de régression assistée par ordinateur: RAO” preprint, 20 Rue Rouget de L'Isle, 94100 Saint Maur, France. The system described runs on a microcomputer and tackles multiple linear regression.Google Scholar
Gale, W A, 1986b. “REX Review” In: Artificial Intelligence and Statistics, Gale, N A, (Ed.) Reading Massachusetts: Addison Wesley pp. 173–228. This was the last and longest of a series of papers on REX. The papers before this do not add much. It describes what REX looked like when it ran, gives an example of the report generated by REX, and describes the inference engine and knowledge representation in detail. It does not discuss the statistical strategy in detail.Google Scholar
Lubinsky, D and Pregibon, D, “Data analysis as search”, Journal of Econometrics, 06, 1988. TESS, the system described, aims to model the results obtained by statisticians in analysing data, not the methods they use. The example studied is regression.Google Scholar
Wolstenholme, D and Neider, J, 1986. “A front end for GLIM” In: Expert Systems in Statistics, Haux, R, ed, Stuttgart and New York: Gustav Fischer. The paper describes early work on GLIMPSE, a front end for the popular statistical system GLIM. The book cited has about ten other papers, but they were written at a planning stage rather than at a stage with systems developed.Google Scholar
Gale, W A, 1986c. “Student phase 1—a report on work in progress” In: Artificial Intelligence and Statistics, Gale, W A, (Ed.), Reading, Massachusetts: Addison Wesley (1986a) pp. 239–266. This described the first version of Student, implemented on a Symbolics Lisp Machine.Google Scholar
Gale, W A, 1987. “Student: a tool for constructing consultation systems in data analysis” In: Proceedings of the 46th Session of the International Statistics Institute, Tokyo. This described the second version of Student, implemented in the new S statistical language.Google Scholar
Gale, W A and Pregibon, D, 1984. “Constructing an expert system for data analysis by working examples” In: COMPSTAT 1984: Proceedings in Computational Statistics, Havranck, T, Sidak, Z and Novak, M, (Eds) Vienna: Physica-Verlag, pp. 227–236. This paper described the motivation and plans for building Student.Google Scholar
Hand, D J, 1987. “A statistical knowledge enhancement system” preprint, Institute of Psychiatry, London. KENS aids a statistician as an on-line reference book.CrossRefGoogle Scholar
Oldford, W and Peters, S, 1986b. “Data analysis networks in DINDE” Proceedings of the Statistical Computing Section, Arlington, Virginia: American Statistical Association, pp. 11–18. This paper describes DINDE's behavior and the knowledge representation behind it.Google Scholar
Becker, R A and Chambers, J M, 1984. S: an Interactive Environment for Data Analysis and Graphics, California: Wadsworth, Belmont. This is the manual for the use of (old) S, a statistical computing system produced by AT&T Bell Laboratories. It contains information on ordering S.Google Scholar
Becker, R A and Chambers, J M, 1986. “Auditing of data analysis” In: Proceedings of Statistical Computation Section, Arlington, Virginia: American Statistical Association, pp. 11–18. The paper describes the use of the (new) S, or QPE, in a task of use to statisticians.Google Scholar
Mallows, C L, 1973. “Some comments on C sub P”, Technometrics 15 pp. 661–667. This paper defined Cp.Google Scholar
Tukey, J W, 1983. “Another look at the future” Heiner, Sachs, and Wilkinson, , (Eds.) Computer Science and Statistics, Proceedings of the Fourteenth Symposium on the Interface, New York: Springer-Verlag pp. 1–8. This paper poses a dozen challenging tasks for statistical computing, some of which may be feasible with K.E methods.Google Scholar
Tukey, J W, 1986. “An alphabet for statisticians' expert systems” In: Artificial Intelligence and Statistics, Gale, W A, (Ed.), Reading, Massachusetts: Addison Wesley, pp. 401–409. This paper appraises the strengths and weaknesses of the idea of applying KE techniques in statistics.Google Scholar
Velleman, P F and Hoaglin, D C, 1981. Applications, Basics and Computing for Exploratory Data Analysis, North Scituate, Massachusetts: Duxbury Press. This book provides practical computer programs and the basic ideas of exploratory data analysis.Google Scholar