https://ir.unisa.ac.za/handle/10500/23894 2026-05-05T15:19:28Z https://ir.unisa.ac.za/handle/10500/24288 Recursive specifications and formal logic. What benefits for intelligent tutoring systems? Tcheeko, L The pedagogical assessment of a tutoring system relies upon a proof of convergence: for such a tutor the correction of student mistakes must not forever delay the teaching process. Such a tutor must provide uniform diagnosis in order to drive the dialog with the student. Student learning can be simulated by a·compilation of knowledge, but it is also necessary to compile knowledge for the teaching process. How can the tutor adjust these levels of compilation while keeping uniform diagnosis? We propose here to use recursive specifications, which consists of formalizing the definition of a class of problems at the same time as their solutions. Such a class "doesn't hide information": this allows its subsequent compilation. 1998-01-01T00:00:00Z https://ir.unisa.ac.za/handle/10500/24287 Specific acquisition of collective belief knowledge for socially motivated multiagent systems Ram, V Shared beliefs and knowledge are an integral part of an organisation's identity and a prerequisite for collective func­tioning. Multiagent systems that support or emulate cooperative problem solving in such a context cannot be constructed without the explicit acquisition and representation of such collective belief knowledge since it is this knowledge that gov­erns coordination, problem decomposition and task allocation. These are complex issues that have a great influence on the overall effectiveness of multiagent systems. However, in much of Distributed Artificial Intelligence (DAI) research and applications, the mechanisms to deal with these issues are directly related to the data abstraction of the problem to be solved or related to the spatial, hierarchical or other structure inherent in the problem. In other words, there is no explicit knowledge acquisition process to identify knowledge for coordination, problem decomposition or task allocation as is the case with domain knowledge in other knowledge-based system development. This paper illustrates the use of congrega­tive cognitive mapping as a technique to elicit and represent collective belief knowledge and shows that it can be used as metaknowledge for coordination, problem decomposition and task allocation. The development of a prototype multiagent system for strategic vigilance is used to illustrate the technique. 1998-01-01T00:00:00Z https://ir.unisa.ac.za/handle/10500/24286 Applied Lambda calculus. Using a type theory based proof assistant Pretorius, L We introduce a number of typed lambda calculi, show how type theory may be used as basis for a proof assistant, and illustrate this with the Coq proof assistant. 1998-01-01T00:00:00Z https://ir.unisa.ac.za/handle/10500/24285 Text categorization as an information retrieval task Paijmans, H A number of methods for feature reduction and feature selection in text classification and information retrieval systems are compared. These include feature sets that are constructed by Latent Semantic Indexing, 'local dictionaries' in the form of the words that score highest in frequency in positive class examples and feature sets that are constructed by relevance feedback strategies such as Rocchio's feedback algorithm or Genetic algorithms. Also, different derivations from the normal Recall and Precision performance indicators are discussed and compared. It was found that categorizers consisting of the words with highest tf .idf values scored best. 1998-01-01T00:00:00Z