This paper is the result of some research in computational stylistics; in particular, the analysis of a document corpus that has attracted the attention of scholars from several disciplines for hundreds of years. This corpus, the Epistles of Saint Ignatius of Antioch, was originally written in Greek but this analysis is of a single translation in English. The analysis has been undertaken using a conventional approach in computational stylistics but has employed a number of contemporary software packages, such as a grammar checker, normally used for text and document creation.
Research in this field predominantly characterises authorship style by the use of document statistics, such as word frequency, sentence and paragraph length and in some cases the recurrence of certain phrases. During the research described here it was considered appropriate to use a grammar checker to identify the existence of a ‘new’ set of characteristics. These include comparing the use of passive voice across the corpus being analysed, the percentage use of prepositions, as well as document statistics such as sentence and paragraph length, and the application of text readability formulas as indicators of writing style.
The corpus analysed in this paper consists of the seven Epistles of Ignatius of Antioch, together with the Epistle of Polycarp to the Philippians. The latter epistle has traditionally been held to authenticate the Ignatian writings. It has been suggested by some church historians that Ignatius was not the author of these epistles and may not in fact, have existed as a person at all. Further, they suggest that two paragraphs in the Polycarp Epistle may have been added later by a second author to authenticate the Ignatian corpus. In order to contribute to the ongoing debate, this paper first examines the Ignatian corpus in order to determine single authorship of the seven epistles. Second, it seeks to determine whether or not the two disputed paragraphs in Polycarp’s Epistle to the Philippians vary in authorship style from the rest of that epistle. Third, it compare authorship style in the two inserted paragraphs of Polycarp’s Epistle with that of the Ignatian corpus in order to make some observations on the hypothesis that a single author was responsible for both.
Whilst change is an inherent characteristic of the IT industry, the difficulty of frequent and timely change in tertiary curricula is a constraint on the ability of universities to adequately meet the requirements of knowledge and expertise expected of new graduates. In this paper, some recently published research concerning the top ten issues for managers of information technology in the USA, Europe and Australia is evaluated in terms of its impact on IS teaching and research. The paper concludes that the top ten issues perceived by IS managers was probably in large part due to change resulting not only from advances in technology but also in response to past failures or inadequacies in the process of delivering high quality information system products to corporate consumers. The need for business and education to be aware of the motivations for change and the constraints that are attendant on it in both environments is emphasised for harmonious progress to prevail in the production and utilisation of new IS graduates.
Petri nets, as a modelling formalism, are utilised for the analysis of processes, whether for explicit understanding, database design or business process re-engineering. The formalism, however, can be represented on a virtual continuum from highly graphical to largely algorithmic. The use and understanding of the formalism will, in part, therefore depend on the resultant complexity and power of the representation and, on the graphical or algorithmic preference of the user. This paper develops a metric which will indicate the graphical or algorithmic tendency of hierarchical coloured Petri nets.
Neural networks (NN) have been intensively used for speech processing. This paper describes a series of experiments on using a single Kohonen Self Organizing Map (KSOM), hierarchically organised KSOM, a backpropagation-type neural network with fuzzy inputs and outputs, and a fuzzy system, for continuous speech recognition. Experiments with different non-linear transformations on the signal before using a KSOM have been done. The results obtained by using different techniques on the case study of phonemes in Bulgarian language are compared.
A Content-Addressable Model of Production Systems, ‘CAMPUS’, has been developed. The main idea is to achieve high execution performance in production systems by exploiting the potential fine-grain data parallelism. The facts and the rules of a production system are uniformly represented as CAM tables. CAMPUS differs from other CAM-inspired models in that it is based on a non-state saving and ‘lazy’ matching algorithm. The production system execution cycle is represented by a small number of associative search operations over the CAM tables which number does not depend, or depends slightly, on the number of the rules and the number of the facts in the production system. The model makes efficient implementation of large production systems on fast CAM possible. An experimental CAMPUS realisation of the production language CLIPS is also reported. The production systems execution time for large number of processed facts is about 1,000 times less than the corresponding CLIPS execution time on a standard computer architecture.
Keywords: associative matching, content-addressable memory (CAM), production systems
This paper describes the creation of a system development methodology suitable for spatial information systems. The concept is substantiated on the fact that spatial systems are similar to information systems in general. The subtle difference being the fact that spatial systems are not yet readily supported by large digital data bases. This fact has diverted attention away from system development to data collection. A spatial system development methodology is derived, based on a historical review of information systems methodologies and the coupling of same with a data collection and integration methodology for the spatially referenced digital data.
Almost by definition, any engineering discipline has quantitative measurement at its foundation. In adopting an engineering approach to software development, the establishment and use of software metrics has therefore seen extensive discussion. The degree to which metrics are actually used, however, particularly in New Zealand, is unclear. Four surveys, conducted over the last eight years, are therefore reviewed in this paper, with a view to determining trends in the use of metrics. According to the findings presented, it would appear that no more than one third of organisations involved in software development utilise software metrics.
Budgetary constraints are placing increasing pressure on project managers to effectively estimate development effort requirements at the earliest opportunity. With the rising impact of automation on commercial software development the attention of researchers developing effort estimation models has recently been focused on functional representations of systems, in response to the assertion that development effort is a function of specification content. A number of such models exist—several, however, have received almost no research or industry attention. Project managers wishing to implement a functional assessment and estimation programme are therefore unlikely to be aware of the various methods or how they compare. This paper therefore attempts to provide this information, as well as forming a basis for the development and improvement of new methods.
Download (gzipped PDF, 252KB)
It is virtually impossible to know everything about any facet of computing as it changes on almost a daily basis. Having said that I believe that it is worth sharing some of the knowledge that I have gained as a result of 5 years of study and experimentation with viruses and virus defense strategies as well as having personally tested nearly 50 anti-virus products.
This paper investigates statistical models for the understanding of the behaviour of scrubweeds in Southland and Otago. Data pertaining to eight scrubweed species have been collected along four transects together with the environmental factors, altitude, slope, aspect and land use classification. Each transect is approximately 80km by 2km, with data being held for every 1ha so that there are approximately 16,000 pixels for each transect. It is important to understand the relationship between the species so that interpolation and extrapolation can be performed. The initial survey, completed in 1992, will be repeated in 1995 and 1998. These surveys will then form the baseline for an understanding of the spread or contraction of the species in farmlands of the South Island. This in turn will assist policy makers in formulating management plans which relate eradication to farmland productivity. This paper deals in detail with one of the transects—Balclutha to Katiki Point.
Keywords: canonical correlation, kriging, log-linear models, logistic regression, spatial correlation, variogram analysis
The paper presents a novel approach towards solving different speech recognition tasks, i.e. phoneme recognition, ambiguous words recognition, continuous speech to text conversion, learning fuzzy rules for language processing. The model uses a standard connectionist system for initial recognition and a connectionist rule-based system for a higher level recognition. The higher level is realised as a Connectionist Fuzzy Production System (CFPS) which makes possible introducing different parameters to the higher level production rules, like: degrees of importance, dynamic sensitivity factors, noise tolerance factors, certainty degrees and reactiveness factors. It provides different approximate chain reasoning techniques. The CFPS helps to solve many of the ambiguities in speech recognition tasks. Experiments on phoneme recognition in the English language are reported. This approach facilitates a connectionist implementation of the whole process of speech recognition (at a low level and at a higher logical level) which used to be performed in hybrid environments. It also facilitates the process of learning fuzzy rules for language processing. All the language processing tasks and subtasks are realised in a homogeneous connectionist environment. This brings all the benefits of connectionist systems to practical applications in the speech recognition area.
A new type of generalised fuzzy rule and generalised fuzzy production system and a corresponding reasoning method are developed. They are implemented in a connectionist architecture and are called connectionist fuzzy production systems. They combine all the features of symbolic AI production systems, fuzzy production systems and connectionist systems. A connectionist method for learning generalised fuzzy productions from raw data is also presented. The main conclusion reached is that connectionist fuzzy production systems are very powerful as fuzzy reasoning machines and they may well inspire new methods of plausible representation of inexact knowledge and new inference techniques for approximate reasoning.
This paper reviews the research and practice of how computer-based output information has been presented in nine different information display formats and the suitability of their use in environments ranging from static, reference-type situations, to complex, dynamic situations. The review while not generating conclusive results suggests that displays are more than a platform to place information. Instead care should be taken to organise, lay out, and pre-process the information so that it enhances the communication between computer and human. The information on the screen should also be designed to augment human cognitive limitations. For instance, human weakness in integrating information across time and multiple sources could be assisted by display formats that integrate the information in the display rather than having the user attempt to integrate that information mentally. If this be the desired outcome, information designers must start to consider performing analyses that help them understand the demands on the human information processing system and hence how information can be presented to augment this weakness. This would have to be further investigated in subsequent research.
Keywords: information display design, output information, visual display formats,
Semantic data models comprise abstractions used, in conceptual database design, to represent real world relationships and aspects of the structure of real world phenomena. Such abstractions have previously been applied to the modelling of spatial concepts, but in the process their semantics are implicitly extended. This paper explicitly extends the semantics of the entity relationship model, defining two specific types of entity set to enable the notion of a thematic layer to be incorporated in entity relationship schemas. It places this in the context of a conceptual modelling framework to be used in the design of spatially referenced databases.
This paper deals with matters relating to toponymy. The concept of indigenous place names is discussed. A view is presented, based on empirical evidence, that current processes for the official recording of names are detrimental to a fair and reasonable representation of indigenous names. Historical events in Aotearoa are examined as well as the existing place name recording process. Research is outlined as to what can be done to examine and redress this situation. A proposition is tendered whereby names can be recorded via a process which is people based and not government based. Research matters surrounding this concept are discussed.
Recently there has been a resurgence of interest in the deductive approach to planning. There are many benefits of this approach but one shortcoming is the difficulty of performing nonlinear planning in this framework. This paper argues that these problems are caused by a flaw in the partial order approach—the lack of structure in such a representation—and proposes an alternative, dynamic programming style approach based on a more structured representation of plans.
The paper presents the major principles of building complex hybrid systems for knowledge engineering where at the centre of the design process is the task of learning (extracting) fuzzy rules from data. An experimental environment FuzzyCOPE, which facilitates this process, is described. It consists of a fuzzy rules extraction module, a neural networks module, module fuzzy inference methods and a production rules module. Such an environment makes possible the use of the three paradigms, i.e. fuzzy rules, neural networks and symbolic production rules, in one system. Automatic rules extraction from data and choosing the most appropriate reasoning mechanism is also provided. Using FuzzyCOPE for building hybrid systems for decision making and speech recognition is discussed and illustrated.
The paper presents a framework of an integrated environment for speech recognition and a methodology of using such environment. The integrated environment includes a signal processing unit, neural networks and fuzzy rule-based systems. Neural networks are used for “blind” pattern recognition of the phonemic labels of the segments of the speech. Fuzzy rules are used for reducing the ambiguities of the correctly recognised phonemic labels, for final recognition of the phonemes, and for language understanding. The fuzzy system part is organised as multi-level, hierarchical structure. As an illustration, a model for phoneme recognition of New Zealand English is developed which exploits the advantages of the integrated environment. The model is illustrated on a small set of phonemes.
The facility to be able to display features of speech in a visual speech aid does not by itself guarantee that the aid will be effective in speech therapy. An effective visual speech aid must provide a visual representation of an utterance from which a judgement on the “goodness” of the utterance can be made. Two things are required for an aid to be effective. Firstly, the clusters of acceptable utterances must be separate from the unacceptable utterances in display space. Secondly, the acoustic features which distinguish acceptable utterances from unacceptable utterances must be evident in the displays of the speech aid. A two part test, called the Visual Display Test (VDT), has been developed to assess a visual speech aid’s capacity to fulfil these requirements.
Download (gzipped PDF, 132KB)