South African Computer Journal - Volume 2003, Issue 30, June 2003

Recently, a Call for Papers has been issued by IFIP's WG 8.2 for the Second Manchester Conference to be held in July 2004.

This paper introduces a grid-based strategy for determining the secret code in the game Mastermind. The grids allow for a decent human player, while also providing a computer player that always has explicit knowledge of the progress

This paper presents an efficient algorithm to classify and retrieve images from large databases in the context of rough set theory. Color and texture are two well-known low-level perceptible features to describe an image's contents used in this paper. The features are extracted, normalized and then the rough set dependency rules are generated directly from the real valued attribute vector. Then the rough set reduction technique is applied to find all reducts of the data which contains the minimal subset of attributes that are associated with a class label for classification. We test three different popular distance measures in this work and find that quadratic distance measures provide the most accurate and perceptually relevant retrievals. The retrieval performance is measured using recall-precision measure, as is standard in all retrieval systems.

A formal specification of a multi-level marketing (MLM) business is presented. Specifying a MLM business boils down to specifying properties of and operations on mathematical forests and trees. The usefulness of the model-based specification language, Z, is investigated as a vehicle for a formal specification of these recursive structures. The specification is presented following a prescribed format, namely the Established Strategy for constructing a Z specification. The Established Strategy is augmented with the notion of proof aimed at corroborating the correctness of critical parts of a specification. We show how attempts at discharging two different proof obligations using the resolution-based, first-order theorem prover OTTER calls for the use of two automated reasoning strategies, namely avoiding equality and using resonance.

In this paper, I present a new family of Commentz-Walter-style multiple-keyword string pattern matching algorithms. The algorithms share a common algorithmic skeleton, which is significantly optimized when compared to the original Commentz- Walter skeleton and subsequently derived improvements. The new skeleton is derived via correctness-preserving stepwise algorithmic improvements, in the Eindhoven style of programming.

Even though the field of pattern matching has been well studied, there are still many interesting algorithms to be discovered. In this paper, we present a new family of single keyword pattern matching algorithms. We begin by deriving a common ancestor algorithm, which naïvely solves the problem. Through a series of correctness preserving predicate strengthenings, and implementation choices, we derive efficient variants of this algorithm. This paper also presents one of the first algorithms which could be used to do a minimal number of match attempts within the input string (by maintaining as much information as possible from each match attempt).

There has been a notable increase recently in the use of temporal logic in new applications, often requiring special and appropriate time structures. The purpose of this paper is to explore ways in which axioms may be used to impose certain properties on the underlying time structure of a temporal logic. We also show why the axiomatisation of properties of the relevant accessibility relations of such a logic needs to be coupled to frames rather than to classes of models.

It has been demonstrated repeatedly that the acoustic models of a speaker-independent speech recognition system can benefit substantially from the application of unsupervised adaptation methods as a means of speaker enrollment. Unsupervised adaptation has however not yet been applied to the statistical language model component of the recognition system. We investigate two techniques with which a first-pass recognition transcription is used to adapt the parameters of the n-gram language model that is used in the recognition search. It is found that best results are achieved when both methods are employed in conjunction with each other. The performance of the adaptation methods were determined experimentally by application to the transcription of a set of lecture speeches. Improvements both in terms of language model perplexity as well as recognition word error-rate were achieved.