Communications of the IIMA


Over the past few decades, most of the existing methods for analysing large growing knowledge bases, particularly Big Data, focus on building algorithms and/or technologies to help the knowledge-bases automatically or semi-automatically extend. Indeed, a vast number of such systems that construct the said large knowledge-bases continuously grow, and most often, they do not contain all of the facts about each process instance or elements that can be found within the process base. As a consequence, the resultant process models tend to be vague or missing value datasets. In view of such challenge, the work in this paper demonstrates that a well-designed information retrieval system or the process mining (PM) methods should present the results or discovered patterns in a formal and structured format qua being interpreted as domain knowledge. To this end, the work introduces a process mining approach that supports further enhancement of existing information systems or knowledge-base through the conceptual means of data analysis. In turn, the paper proposes a semantic-based process mining and analysis method, or better still, information retrieval and extraction system - that is capable of detecting patterns or unobserved behaviours within any given knowledge base by making use of the underlying semantics or properties (metadata) that describes the available data. Thus, the proposed approach is grounded on the semantic modelling and process mining techniques. The work illustrates this method using the case study of Learning Process. The goal is to discover user interaction patterns within a learning execution environment and respond by making decisions based on the semantical analysis of the captured users data. Practically, the method applies semantic annotation and ontological representation of the learning process domain data and the resultant models in order to discover patterns automatically by means of semantic reasoning. Theoretically, the process mining and modelling method show that a way of addressing the common challenge with computational intelligent systems or methods is through an effectively well-designed and fit for purpose system that meets the requirements and needs of the intended users. In other words, this paper applies effective reasoning methods to make inferences over a process knowledge-base (e.g. learning process) that leads to an automated discovery of learning patterns or behaviour.