(448d) Semantic Repository for Biorefining Model Integration

The increasing availability of biorefining models and the heterogeneity characterising them necessitates efficient acquisition, discovery and integration tools to enable their reuse. This has been addressed by the CAPE-OPEN standard which significantly facilitates model reusability and interoperability (Braunschweig et al., 2004). Still, efficiency and flexibility of model discovery remain a challenge. This is primarily because the model interface semantics are implicit, understandable by humans, but difficult to allow for automation controlled by machines. In order to enhance model discovery and subsequently model integration, we propose a semantic repository biorefining models benefiting from a semantic annotation and implemented as a semantic web service.

The semantic annotation in the form of a full description of models is accomplished by the biorefining domain ontology developed in Web Ontology Language (Koo et. al 2017), hence allowing for a precise description of model functionality, their inputs, outputs, precondition and effects (IOPE). Models are stored as instances of ontology top level classes Model By Functionality, Model By Input, Model By Output, Model By Platform, or their respective subclasses.

Model integration, including model discovery is enabled by introducing an established the Web Ontology Language Semantic Mark-Up for Services (OWL-S) framework (Martin et. al 2004). In consequence, the models are recorded as web services and described by OWL-S which uses three interconnected sub-ontologies (Profile.owl, Process.owl, Service.owl), each linked to a model instance in the biorefining domain ontology.

The proposed model repository operates by following five stages:

i) Model Registration for registering a model and hence providing all IOPE-s for that model; the registered model becomes an instance of the biorefining ontology;

ii) Model Publishing for sharing all registered model;

iii) Model Discovery for discovering other models and concomitantly integrating it with other models; model discovery is enabled by an I/O semantic matchmaking with I inputs a requesting model. Candidate models are ranked by the level of semantic similarity with a full explanation enabling the user an informed selection. Increased flexibility and a wider range of options are achieved by the introduction of partial matching, which, again, are fully described for their relevance;

iv) Model Selection, the stage where the user chooses the most appropriate candidate model; offered for integrating it with requesting model;

v) Model composition triggers the synchronization algorithm for execution and data exchange between the selected models.

The functionality of the proposed semantic repository was tested using several biorefining related scenarios, i.e. modelling of bioethanol production and modelling several bioprocesses.

References

Braunschweig, B., Fraga, E., Guessoum, Z., Marquardt, W., Nadjemi, O., Paen, D., Pinol1, D., Roux, P. Sama, S., Serra, M., Stalker, I.,Yang, A., 2004. CAPE web services: the COGents way, Comput. Aided Chem. Eng., 18 (C) , pp. 1021–1026

Koo, L., Trokanas, N., and Cecelja, F. , 2017. A semantic framework for enabling model integration for biorefining. Computers & Chemical Engineering. http://doi.org/10.1016/j.compchemeng.2017.02.004

Martin, D., Burstein, M., Lassila, O., Paolucci, M., Payne, T. and McIlraith, S., 2004. Describing Web Services using OWL-S and WSDL. http://www.daml.org/services/owl-s/1.1/owl-s-wsdl.html