Abstract
In consideration of the growing number of models and data distributed by various computer aided process engineering (CAPE) tools with various modelling assumptions and methods resulted in heterogeneity and they are prone to remain implicit to the modellers who created them. Therefore, this amplifies the problem of not only redundant work and further limits the potential of reusability of these models and data but also potential establishment of interoperability. For this reason, the importance of developing a better management of information regarding model and data in the CAPE community is recognised. In response to these challenges, the aim of this research is to contribute to the development of a decision support framework to manage the large number of available models and data and its heterogeneity, with a goal to enable interoperability between process models and/or data. The new techniques benefiting from semantic technology is introduced as a solution to provide a unified semantic framework to capture knowledge of the models and data in the domain of biorefining, and to promote reuse of these existing models and data. In addition, semantic modelling is used to depict all relevant concepts in an ontology by capturing the associations, ensuring the understanding of the exchanged knowledge during models and data interoperation. To enable automatic discovery of models and data to further support model and/or data integration, semantic algorithms are proposed as methods to quantify semantic relevance between different models and data. Such methodology is introduced as input/output matching that is a combination of the vector space modelling and graph modelling, whilst allowing partial matching. In vector space modelling, the properties of inputs and outputs of the model, and outputs of the data are converted into vectors and compared against the matching criteria. In graph modelling, the semantic captured in ontology is presented as a single graph to compute distance measurement to calculate similarity between concepts in the graph in order to propose a model or data to be integrated. This approach concomitantly supports the composition of the complex chains of models and data by recursive repetition of the backward matching process. Overall, this thesis presents a step forward toward a more flexible approach to the integration among the process models and data by improving search for decision making process using the application of semantic models, which are increasingly used.