Computer science senior section.

* Jones, R., Departments of Physical Science, Emporia State University. Chaining Case-Based Reasoners. Case-based reasoners (CBRs) contain a set of stored records or "cases" which are compared with new inputs ("inquiries"). If the inquiry closely resembles one of the stored records the output (prediction) associated (stored) with that case is assumed to apply to the new situation as well (Case-Based Reasoning, J. Kolodner, Morgan Kaufmann, 1993). Traditional CBRs follow a radical behaviorist-like model in that only inputs and outputs are employed; there are no intermediate ("internal" or "hidden") variables. All reasoning goes straight from inputs to conclusions. In our new system, however, input variables first generate intermediate results and these, in turn, are then combined (chained, like logic-based production systems) in order to calculate further intermediate variables and, ultimately, make output predictions. Any number of CBRs may then be combined (chained) to produce a complete fuzzy knowledge based expert system. In effect, each rule in a production system has been replaced by a case-based reasoner.

* Kumaran, J. K. Mitchell, and A. van de Liefvoort. School of Computing and Engineering, University of Missouri-Kansas City. An Analytic Model Of Correlations Induced In A Packet Stream By Background Traffic In {Ip} Access Networks. In this paper we show that background traffic at a server in an IP access network can have a significant effect on the performance for a tagged class. In particular, we present an analytic model for two traffic classes with general arrival patterns and general service requirements that are multiplexed at an IP server. The model is general in that it incorporates first- and second-order statistics for the processes involved and is not limited to a heavy traffic analysis. We study the effects of multiplexing on the departure process for the tagged class, and quantify the impact of the background traffic on the mean and variance. More importantly, it shows that the background traffic can introduce slowly decaying correlations in the tagged stream, which, if ignored, will cause performance predictions to be underestimated by several orders of magnitude. We show that significant correlations are introduced by the background traffic when the arrival time distribution for the background class deviates from those of the tagged class, thereby severely limiting the operational range of the server. Supported in part by NSF under grant No. ANI 0106640

* Naugler, R. Department of Computer Science, Southeast Missouri State University. Functional Programming In Java And C#. In Scheme, ML and other functional languages functions are "first-class" objects and can be passed as parameters, constructed in other functions and returned as values. This allows a distinct programming style quite different from the style in imperative and most object oriented languages. Superfically neither Java nor C# has functions, although methods and static methods serve well. In both Java and C# it is fairly easy to mimic some of the important features of functional programming. Javas and C# make an interesting contrast since different constructs in each must be used to do this. In Java interfaces and inner classes can be used. C# does not have inner classes. In C# delegates are used to implement these features instead of interfaces and inner classes. It is shown how to program in a functional way in both Java and C#. The constrasting way this is done shed light on the similarities and the differences between the interface and inner class combination in Java and delegates in C#. Functional languages are much superior for teaching functional programming. However, these approaches allow modes of thinking learned in the study of functional languages to be used directly in programmes written in Java and C#, and they provide a comparison/contast of useful and often poorly understood constructs in Java and C#.

* Saquer, J. M. Computer Science Department, Southwest Missouri State University. Formal Concept Analysis And Data Mining. Formal concept analysis (FCA) is a branch of mathematics that emerged when a group of researches were trying to develop applications for lattice theory. It formalizes the notion of a concept in a given context. For example, consider the context of transactions at a grocery store where each transaction consists of the items bought together. A concept here is a pair of two sets (X, Y). X is the set of all transactions that contain all the items in Y and Y is the set of items common to all the transactions in X. A successful area of application for FCA has been data mining (DM), which deals with the automatic extraction of useful patterns and knowledge from large volumes of data. Data mining is a new interdisciplinary topic in computer science that ties ideas from other well established fields such as databases, machine learning, algorithms, information retrieval, pattern recognition, and computer vision. DM has been used successfully in fraud detection, bioinformatics, sequence analysis, and predicting users' behaviors on the World Wide Web. In this presentation, I will give a brief overview of the basic notions of FCA. I will also show how the idea of a formal concept in FCA can be successfully used in many areas of data mining including clustering and generating association rules. The ideas I will show can be easily adopted in an upper-level undergraduate and low level graduate course on data mining.

* Shade, E. Computer Science Department, Southwest Missouri State University. A Parametric Automata-Theoretic Programming Language For The Chomsky Hierarchy. The Chomsky hierarchy includes the regular, context-free, context-sensitive, and unrestricted formal languages, which have both grammatical and automata-theoretic definitions. While this makes a theoretical comparison of the descriptive power of these languages possible, many people (particularly students) find it difficult to understand the distinctions in intuitive and practical terms. I have defined a programming language with one conditional statement, one iterative statement, nonrecursive subroutines, and a nondeterministic guess operator that can be parameterized to correspond precisely to each of the formal languages in the Chomsky hierarchy. For example, a language is context-free if and only if it can be recognized by a program that is permitted to use the push, pop, top, and empty stack operations. Programs are usually quite short and easier to understand than either grammars or automata.

* Wang, Y. Department of Computer Science, Southwest Missouri State University. Pedagogy For Introducing The Concept Of Object-Oriented Programming With Java In Cs I. Object-oriented programming techniques are essential to contemporary software development and the computer industry. In order to expose the students to the concept of object-oriented programming as early as possible, many schools have chosen using Java in the course CS I. However, the numerous Java textbooks have shown that there is a broad range of opinions on how and in what sequence the basic programming schemes and object-oriented programming techniques should be taught to those first-time programmers. In this paper the author compares those different teaching pedagogies, and intends to stir up a discussion on this issue among the peer educators.

David R. Naugler

Southeast Missouri State University