Expert systems (Computer science)

The subject of this thesis is the design and
implement at ion of an expert system from a standard data
base management software language. The advantages and
limitations of such a system design are discussed and
supported by an accompanying implementation. Both, the
design and implementation, demonstrate what gives the
expertise or personification of human reasoning to a
machine and why this type of reasoning is well suited to
certain types of problems. This fundamental departure from
traditional deterministic analytical problem solving is
accomplished by developing a system that is heuristic in
nature. This heuristic implementation provides for a
system that assists in the development of an emerging
solution, rather than a deterministic solution in and of
itself (i.e., a system that is programmed with a set of
meta-knowledge rules that governs the decision making
process and acts upon a second set of knowledge rules).

The Knowledge Based System CARAT is developed to rate any existing steel truss bridge. CARAT is capable of rating statically determinate and indeterminate truss bridges. Corrosion, temperature effects and the remaining fatigue life of the truss members can be evaluated using CARAT. The other expert system SKEWRAT is developed to determine the strength of skew bridges. Skew AASHTO girder and slab bridges can be analyzed using SKEWRAT. The architecture of both CARAT and SKEWRAT are based on production system model and set of IF-THEN rules are developed using EXSYS Professional Shell editor. The inference mechanism fire rules according to the built-in reasoning process. The suggestions given by Bakht and Jaeger (8) are incorporated to determine the strength characteristics of skewed AASHTO girder and slab bridges. Programming Languages FORTRAN and C are used extensively for the software development. The validity of both the softwares are verified and illustrated in detail with four examples.

The automation of bridge strength evaluation by utilizing artificial intelligence tools is presented herein. The study involved the development of a microcomputer based expert system, REX (Rating EXpert). REX is an interactive menu-driven system combining the expert system shell EXSYS, the Grillage program 'BRIDGES', and various other pre- and post- processing devices. The system is capable of analyzing and assessing the load carrying capacity of solid and voided slab, and AASHTO girder and slab bridges.

This thesis describes the development of a microcomputer based prototype expert system, RETAININGEARTH, for the selection and design of earth retaining structures. RETAININGEARTH is an interactive menu-driven system and consists of two modules--the selection module, SELECTWALL and the design module. SELECTWALL is developed using the rule-based M.1 knowledge engineering shell and it makes a choice of the most appropriate retaining structure from a list of ten typical walls. The design module consists of five independent design programs which performs detailed designs of the concrete gravity and cantilever walls, gabions, reinforced earth and sheetpile structures. The SELECTWALL and the design module are linked by the M.1 external code EXT through a control program CALL. All the design procedures are coded using the C programming language.

The intent of this thesis is to show how rule structures can be derived from influence diagrams and how these structures can be mapped to existing rule-based shell paradigms. We shall demonstrate this mapping with an existing shell having the Evidence (E) --> Hypothesis (H), Certainty Factor (CF) paradigm structure. Influence diagrams are graphical representations of hypothesis to evidence, directed forms of Bayesian influence networks. These allow for inferencing about both diagnostic and predictive (or causal) behavior based on uncertain evidence. We show how this can be implemented through a Probability (P) to CF mapping algorithm and a rule-set conflict resolution methodology. The thesis contains a discussion about the application of probabilistic semantics from Bayesian networks and of decision theory, to derive qualitative assertions about the likelihood of an occurrence; the sensitivity of a conclusion; and other indicators of usefulness. We show an example of this type of capability by the addition of a probability range function for the premise clause in our shell's rule structure.

This thesis presents a procedure for the selection and design of retaining walls using an expert system. The selection part is formulated in the form of production rules by OPS5, a programming language for production systems, and the design part is written in the procedural language, BASIC. Nine different types of retaining walls are incorporated in the knowledge base of the selection part, and three types of walls in the design part of the expert system. The selection and design parts are combined using OPS5 support routines.

The problem of finding optimal paths for a robot navigating in an environment where the position of each obstacle is precisely known has received much attention in the literature, however, the majority of applications problems for a robot would require it to navigate in a completely unknown. This paper focuses on an approach to solving the problem of robot navigation in an unknown, unstructured, two-dimensional environment where the positions of the polygonal obstacles were fixed in time. Few studies have reported on the utilization of an expert system to govern robot motion. This study relied on a knowledge-based expert system that interacted with lower-level procedures to carry out path finding and exploration functions. The expert-system shell used was OPS5 which ran on top of Lisp.

This thesis expands upon an existing noise cleansing technique, polishing, enabling it to be used in the Software Quality Prediction domain, as well as any other domain where the data contains continuous values, as opposed to categorical data for which the technique was originally designed. The procedure is applied to a real world dataset with real (as opposed to injected) noise as determined by an expert in the domain. This, in combination with expert assessment of the changes made to the data, provides not only a more realistic dataset than one in which the noise (or even the entire dataset) is artificial, but also a better understanding of whether the procedure is successful in cleansing the data. Lastly, this thesis provides a more in-depth view of the process than previously available, in that it gives results for different parameters and classifier building techniques. This allows the reader to gain a better understanding of the significance of both model generation and parameter selection.

This thesis presents a noise handling technique that attempts to improve the quality of training data for classification purposes by eliminating instances that are likely to be noise. Our approach uses twenty five different classification techniques to create an ensemble of classifiers that acts as a noise filter on real-world software measurement datasets. Using a relatively large number of base-level classifiers for the ensemble-classifier filter facilitates in achieving the desired level of noise removal conservativeness with several possible levels of filtering. It also provides a higher degree of confidence in the noise elimination procedure as the results are less likely to get influenced by (possible) inappropriate learning bias of a few algorithms with twenty five base-level classifiers than with a relatively smaller number of base-level classifiers. Empirical case studies of two different high assurance software projects demonstrate the effectiveness of our noise elimination approach by the significant improvement achieved in classification accuracies at various levels of filtering.

During the last few years, Service-Oriented Architecture (SOA) has been considered to be the new phase in the evolution of distributed enterprise applications. Even though there is a common acceptance of this concept, a real problem hinders the widespread use of SOA : A methodology to design and build secure service-oriented applications is needed. In this dissertation, we design a novel process to secure service-oriented applications. Our contribution is original not only because it applies the MDA approach to the design of service-oriented applications but also because it allows their securing by dynamically applying security patterns throughout the whole process. Security patterns capture security knowledge and describe security mechanisms. In our process, we present a structured map of security patterns for SOA and web services and its corresponding catalog. At the different steps of a software lifecycle, the architect or designer needs to make some security decisions.