South African Computer Journal - Volume 2008, Issue 40, 2008
Volumes & issues
Volume 2008, Issue 40, 2008
Author Dewald RoodeSource: South African Computer Journal 2008, pp 1 –2 (2008)More Less
In a recent editorial in the Information Systems Journal  Emmanuel Monod and Richard Boland, Guest Editors of a special ISJ issue on Philosophy and Epistemology, argued: "It is time for IS to stop repeating its history and that of classical physics. To repeat that history is to continue reproducing the mistakes of others. Unless we break from our history, we will not grow as a discipline." (Emphasis added.)
Source: South African Computer Journal 2008 (2008)More Less
The current issue contains a special section with selected papers based on submissions to the 18th annual conference of the Pattern Recognition Association of South Africa. Every year, this conference brings together South African researchers as well as visitors from other countries with an interest in pattern recognition and related subjects.
Evaluating and improving morpho-syntactic classification over multiple corpora using pre-trained, "off-the-shelf", parts-of-speech tagging tools : reviewed articleSource: South African Computer Journal 2008, pp 4 –10 (2008)More Less
This paper evaluates six commonly available parts-of-speech tagging tools over corpora other than those upon which they were originally trained. In particular this investigation measures the performance of the selected tools over varying styles and genres of text without retraining, under the assumption that domain specific training data is not always available. An investigation is performed to determine whether improved results can be achieved by combining the set of tagging tools into ensembles that use voting schemes to determine the best tag for each word. It is found that while accuracy drops due to non-domain specific training, and tag-mapping between corpora, accuracy remains very high, with the support vector machine-based tagger, and the decision tree-based tagger performing best over different corpora. It is also found that an ensemble containing a support vector machine-based tagger, a probabilistic tagger, a decision-tree based tagger and a rule-based tagger produces the largest increase in accuracy and the largest reduction in error across different corpora, using the Precision-Recall voting scheme.
Shape from shading using MRF optimization with Gibbs sampling with quadruplet cliques : pattern recognition special editionSource: South African Computer Journal 2008, pp 11 –17 (2008)More Less
This paper extends the MRF formulation approach developed solving the shape from shading problem. Our method extends the Gibbs sampling approach to solve an MRF formulation which characterizes the Shape from Shading (SFS) problem under Lambertian reflectance conditions (the algorithm is extensible to other lighting models). Our method uses a simpler set of energy functions (on point quadruplets), which is faster to converge, but less accurate.
Source: South African Computer Journal 2008, pp 18 –22 (2008)More Less
We explore the construction of a system to classify the dominant emotion in spoken utterances, in a environment where resources such as labelled utterances are scarce. The research addresses two issues relevant to detecting emotion in speech: (a) compensating for the lack of resources and (b) finding features of speech which best characterise emotional expression in the cultural environment being studied (South African telephone speech). Emotional speech was divided into three classes: active, neutral and passive emotion. An emotional speech corpus was created by naive annotators using recordings of telephone speech from a customer service call centre. Features were extracted from the emotional speech samples and the most suitable features selected by sequential forward selection (SFS). A consistency check was performed to compensate for the lack of experienced annotators and emotional speech samples. The classification rate achieved is 76.9%, with a 95% classification rate for active emotion.
Source: South African Computer Journal 2008, pp 23 –30 (2008)More Less
Retinal fundus images are used in the diagnosis and treatment of many eye conditions, such as diabetic retinopathy and glaucoma. During a clinical examination, an ophthalmologist is able to determine the onset of disease by taking certain features of the retinal vessels of the fundus into account. Often the ophthalmologist will need to select what parts of a retinal fundus image constitute vessels, so that certain statistics such as the thickness of the vessels can be calculated. Labelling all the vessels however, is a tedious and time consuming process. We make use of an edge detection approach, followed by connected component analysis and a fast Shi-Karl level-set technique, to extract the vessels from the image. Our work focuses on using image processing techniques in order to develop a computer program that can automatically and interactively detect and segment blood vessels in these images, thereby saving the ophthalmologist considerable time. We create our tool as a series of plug-ins for ImageJ, which is a public domain, Java-based image processing program developed at the National Institutes of Health. By doing so, we facilitate the use of our tool as part of a bigger system. Furthermore, ophthalmologists can easily modify the proposed vessel segmentation of our system using ImageJ, and have an ever growing library of image processing plug-ins at their disposal, to annotate, enhance and measure images.
Source: South African Computer Journal 2008, pp 31 –36 (2008)More Less
Land surface temperatures (LSTs) can be approximated from brightness temperatures observed from satellites. Estimation errors between observed brightness temperatures and a brightness temperature model of a given pixel would provide information for a pixel concerned. Robust fitting of an observed Diurnal Temperature Cycle (DTC) taken over a day of a given pixel without cloud cover and other abnormal conditions such as fire can be used to derive a data-based brightness temperature model for a given pixel. A novel Singular Value Decomposition (SVD) method and an improved Reproducing Kernel Hilbert Space (RKHS) interpolator method (using a robust estimation method to approximate the coefficients of the RKHS interpolator) are proposed and compared to a pseudo-physical model approach. In this paper, diurnal brightness temperatures from the METEOSAT Second Generation (MSG) satellite were used to obtain a model. The simulation results show that the approach based on SVD outperforms other approaches in the sense that an accurate model which can be successfully used for the interpolation of missing diurnal cycle data, is more often found.
Naive Bayesian classifiers for multinomial features : a theoretical analysis : pattern recognition special editionSource: South African Computer Journal 2008, pp 37 –43 (2008)More Less
We investigate the use of naive Bayesian classifiers for multinomial feature spaces and derive error estimates for these classifiers. The error analysis is done by developing a mathematical model to estimate the probability density functions for all multinomial likelihood functions describing different classes. We also develop a simplified method to account for the correlation between multinomial variables. With accurate estimates for the distributions of all the likelihood functions, we are able to calculate classification error estimates for any such multinomial likelihood classifier. This error estimate can be used for feature selection, since it is easy to predict the effect that different features have on the error rate performance.
Source: South African Computer Journal 2008, pp 44 –50 (2008)More Less
The durations of phonemes varies for different speakers. To this end, the correlations between phonemes across different speakers are studied and a novel approach to predict unknown phoneme durations from the values of known phoneme durations for a particular speaker are presented, based on the maximum likelihood criterion. Several interesting patterns are observed. Phonemes from the same broad phonetic class tend to covary most strongly (and therefore intra-class predictions of unknown phoneme durations are most accurate), but significant cross-class correlations are also present. Consequently, knowledge of only a few highly-correlated phonemes' durations is necessary to make a good duration prediction.
Author Scott HazelhurstSource: South African Computer Journal 2008, pp 51 –62 (2008)More Less
Understanding which genes are active, and when and why, is an important question for molecular biology. Expressed Sequence Tags (ESTs) are a technology used to explore the transcriptome (a record of this gene activity). ESTs are short fragments of DNA created in the laboratory from mRNA extracted from a cell. The key computational step in their processing is clustering: putting all ESTs associated from the same RNA together. Accurate clustering is quadratic in time in average EST length and number of ESTs, which makes naïve algorithms infeasible for real data sets. The wcd EST clustering system is an open source clustering system that provides efficient implementations of key distance measures, heuristics for speeding up clustering, a pre-clustering booster based on suffix arrays, as well as parallelised implementations based on MPI and Pthreads. This paper presents the underlying algorithms in wcd. The code is available from http://code.google.com/p/wcdest.
Source: South African Computer Journal 2008, pp 63 –73 (2008)More Less
Mobile communications have had a dramatic impact on our society. Users take for granted the ability to make or receive telephone calls from almost any location, at any time. This increase in mobility and convenience to users has also burdened them with new reachability and time-management issues. In this paper, we present the results of a survey which evaluates the current mobile perception and practice of users, focusing on their call-management behaviour and their need for such a system. We discuss the results and hypothesise on future developments in this area.
Source: South African Computer Journal 2008, pp 74 –82 (2008)More Less
The implementation of Enterprise Information Systems generally has disruptive implications for the workforce directly affected by them. Normal change management procedures typically address such issues through user training programmes, based on the perhaps unfounded assumption that users just need to understand the new facilities offered to be able to adapt to new ways of working. It is assumed that any emotional distress and unhappiness would disappear as soon as the users realize the benefits of the new system. In this paper we report on an investigation of such a situation where an ERP system replaced a home-grown student system that had been in use for many years. An inductive analysis of interview data was undertaken, leading to a framework of five linked categories. We draw conclusions from the framework which point towards more effective ways an organisation can deal with the affective responses of users, mitigating subsequent negative job impact.
A feasibility study on the use of 'smart' pens in South African teaching and learning environments : reviewed articleSource: South African Computer Journal 2008, pp 83 –94 (2008)More Less
Interactive whiteboards are being installed in many schools in developed and even in developing countries. As they are currently quite expensive, the Eastern Cape Department of Education (ECDoE) initiated several studies to explore the various types of interactive whiteboard systems. This article describes one of the feasibility studies which investigated the use of a 'smart' pen interactive whiteboard technology in previously disadvantaged South African schools as a means for promoting learning in the classroom while simultaneously developing ICT skills. Most investigations of interactive whiteboards review the use of the more expensive 'smart' board technology. This article details the five most common technical problems experienced by teachers and learners of the 'smart' pen interactive whiteboard technology: calibration and infrastructural issues; hardware and software; training and support; timetabling; and portability. This article then compares the findings with the 'smart' pen with previous research that investigated the more expensive 'smart' boards. The findings reveal that although the types of interactive whiteboard systems have slightly different benefits and drawbacks, the critical issues seem not lie in the choice of the type of interactive whiteboard technology per se, but rather in the way in which they are deployed in previously disadvantaged schools. The key issue seems to relate to teachers' prerequisite ICT literacy and integration skills that need to be in place prior to the installation of interactive whiteboard technology. It would seem that unless teachers are sufficiently ICT literate and the school is in a position to support the use of interactive whiteboard technology, the interactive whiteboards are not used optimally or in a sound pedagogical manner in the classroom.
Author Stefan GrunerSource: South African Computer Journal 2008, pp 95 –96 (2008)More Less
40 Years ago: 1968! While Woodstock-inspired hippies, with long hair, made love, not war, and while leftwing students, successfully rebelled for a more liberal society in general and against the academic establishment in particular, an event took place almost unnoticed, in a small Bavarian town at the foot of the Alps, which -in retrospective- turned out to be as important for our profession as the students' rebellion and the "Spirit of 68" was for the general liberalisation and opening up of the Western societies.