Article

Medidas de Complejidad Cuantitativas para Sistemas Expertos Basados en Reglas

Inteligencia artificial: Revista Iberoamericana de Inteligencia Artificial, ISSN 1137-3601, Vol. 13, Nº. 43, 2009, pags. 16-31 01/2009;
Source: OAI
0 Bookmarks
 · 
87 Views
  • [Show abstract] [Hide abstract]
    ABSTRACT: The development and use of knowledge-based (expert) systems has grown dramatically across a broad range of industries. Yet despite its growing importance, the study of expert systems lacks a cohesive framework for differentiating and comparing expert systems initiatives across different applications and in different industrial settings. The problem for IS managers is that a system that works in one situation may ot be appropriate for another. This article presents a classification methodology for the systematic evaluation of a broad range of expert systems. Of primary concern in this study is the measurement of the complexity of such systems. Complexity in the area of expert systems consists of two basic dimensions. The first dimension is the complexity of the underlying knowledge residing with the key experts. The second dimension of the framework focuses on the complexity of the technology incorporated into a given system. This framework is then applied to a sample of 50 successfully developed knowledge-based systems. The results can be used as a foundation for generating research hypotheses and for development time, budget, staffing, organizational control, and organizational participation.
    MIS Quarterly. 01/1991; 15:455-472.
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Four measures are developed and empirically tested that attempt to quantify the complexity of rule-based Prolog programs in terms of measuring the dependency between rules. The measures are lines of code (LOC), number of rules (NR), number of unique predicate nodes (UPN), and McCabe's cyclomatic complexity model, v(G). The UPN and v(G) measures are developed by reducing a Prolog program to its most abstract graphical form and then counting the nodes and arcs represented by the graph. The four measures are tested in terms of their ability to distinguish between a set of 80 professionally developed programs, divided into groups of error-prone and error-free programs. Unpaired student's t-tests showed that UPN is a significantly better measure of complexity (ρ=0.002–0.006) than any of the other measures (ρ=0.096–0.609). Finally, by plotting the cumulative frequency of errors across the sample, it is possible to observe a threshold point of around 35 ± 5 UPN, above which Prolog programs contain significantly more errors (ρ=0.000).
    Journal of Systems and Software. 01/1998;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Knowledge-based engineering and computational intelligence are expected to become core technologies in the design and manufacturing for the next generation of space exploration missions. The literature is contradictory on how we are to assess such systems. Studies indicate significant disagreement regarding the amount of testing needed for system assessment. The sizes of standard black-box test suites are impractically large since the black-box approach neglects the internal structure of knowledge-based systems. On the contrary, practical results repeatedly indicate that only a few tests are needed to sample the range of behaviors of a knowledge-based program. In this paper, we model testing as a search process over the internal state space of the knowledge-based system. When comparing different test suites, the test suite that examines larger portion of the state space is considered more complete. Our goal is to investigate the trade-off between the completeness criterion and the size of test suites. The results of testing experiment on tens of thousands of mutants of real-world knowledge based systems indicate that a very limited gain in completeness can be achieved through prolonged testing. The use of simple (or random) search strategies for testing appears to be as powerful as testing by more thorough search algorithms.
    International Journal of Artificial Intelligence Tools 01/2000; 9:153-172. · 0.25 Impact Factor

Full-text

View
0 Downloads
Available from