Masters Degrees (Information Science)

Permanent URI for this collection

https://scholar.sun.ac.za/handle/10019.1/628

Browse

Browsing Masters Degrees (Information Science) by browse.metadata.advisor "Britz, Katarina"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Repairing classical ontologies using defeasible reasoning techniques
    (Stellenbosch : Stellenbosch University, 2021-03) Coetzer, Simone; Britz, Katarina; Stellenbosch University. Faculty of Arts and Social Sciences. Dept. of Information Science.
    ENGLISH SUMMARY : Ontologies provide knowledge engineers with the ability to represent and encode knowledge in a formal language so that it can be reasoned over by a computer. Notable benets include the ability to source new knowledge by making statements that are implicitly deduced explicitly available to the end-user, to classify individuals or instances and to check the addition of new knowledge for logical consistency. Given the nature and goal of ontologies, a successful application of ontologies relies on (1) representing as much accurate and relevant domain knowledge as possible, (2) while maintaining logical consistency. As the successful implementation of a real-world ontology is likely to contain many concepts and intricate relationships between the concepts, it is necessary to follow a methodology for debugging and rening the ontology. A myriad of ontology debugging approaches (some of them instantiated in tools) have been developed to help the knowledge engineer pinpoint the cause of logical inconsistencies and rectify them in a strategic way. Rodler (2015) and Schekotihin et al. (2018) build out the ontology debugging basics by introducing an interactive ontology debugging methodology: this interactive ontology debugging framework, which has also been implemented as a Prot eg e plug-in, OntoDebug, methodically and iteratively asks users queries to narrow down the inconsistency to just one diagnosis, at which time the user can make a more informed decision about how to repair the diagnosis. This approach guides the user in the debugging process. We show however that this approach can sometimes lead to unintuitive results, which may then lead the knowledge engineer to opt for deleting potentially crucial and nuanced knowledge. This is due to the focus of the interactive ontology debugging approach to be on classical, monotonic knowledge bases { and indeed, in the classical/ monotonic sense, it is only by deletion, not extension of the knowledge base, that coherence can be obtained. However, it may at times be desirable to deal with the unintuitive results produced by weakening rather than deleting faulty axioms. We provide a methodological and design foundation for weakening faulty axioms in a strategic way using defeasible reasoning tools. Our methodology draws from Rodler's (2015) interactive ontology debugging approach which not only localises faulty axioms but provides the knowledge engineer with a strategic way of resolving them by presenting the root cause inconsistencies rst. We extend this approach by creating a methodology to systematically nd con ict resolution recommendations. Importantly, our goal is not to convert a classical ontology to a defeasible ontology { therefore we do not use defeasible reasoning support through, for example, the computation of rational closure. Rather, we use the denition of exceptionality of a concept, which is central to the semantics of defeasible description logics, and the associated algorithm (as can be found in Britz et al. 2019) to determine the extent of a concept's exceptionality (their ranking); then, starting with the statements containing the most general concepts (the least exceptional concepts) weakened versions of the original statements are constructed; this is done until all inconsistencies have been resolved.
  • Thumbnail Image
    Item
    Visual exploration of large geolocation-rich data sets using formal concept analysis
    (Stellenbosch : Stellenbosch University, 2022-04) du Toit, Tiaan; Britz, Katarina; Fischer, Bernd; Stellenbosch University. Faculty of Arts and Social Sciences. Dept. of Information Science.
    ENGLISH SUMMARY: The rate at which data is being generated is ever-increasing, resulting in an abundance of very large data sets of di erent structures, all requiring improved methods of data capture, processing and storage. This in turn requires improved representation and data exploration methods. In an age of web applications, one such structure that has gained popularity is semi-structured data. Unlike relational data which can be speci cally queried, semi-structured data relies on di erent methods for data exploration and knowledge discovery. One such method is the visualisation of data beyond a purely granular textual format, in a dynamic and reactive manner. ConceptCloud, a exible interactive web application for exploring, visualising, and analysing semi-structured data sets, uses a combination of an intuitive tag cloud visualisation with an underlying formal concept lattice to provide a formal structure for navigation through a data set. It is an e ective, robust and scalable tool that allows for extension. The underlying formal concept lattice also allows for alternative possibilities of exploring data. This research describes the development and implementation of extensions made to the existing ConceptCloud tool, which are focused on improving the visualisation of, and the knowledge discovery through interaction with a data set, especially data with aspects that could be visualised in a more e ective manner. These extensions include a map based viewer for visualising geolocation data, a graph based viewer for visualising the composition of the data as well as a REST API to allow for mobile application development and further unique visualisations. These extensions are demonstrated and evaluated by visualising and exploring two semi-structured data sets in the viticultural domain, namely atmospheric measurements of grape growing regions and wine reviews. It is shown how these extensions aid in and support data exploration and knowledge discovery of these multi- faceted semi-structured data sets. These visualisations can not only be re ned and expanded upon to include other types of visualisations, but also improve data mining and knowledge discovery using ConceptCloud. This could result in further research and improvements in similar tools and processes.