It’s Linh Hoang, the intern from project 3. This week, along with our supervisors, my co-intern and I tried to finalize the YW model with our own vocabulary. We also created an UML diagram to represent the model. This is a good way for outsiders to conceptualize and understand what are the components of YW model and how they connect to each other. Even though the model is not perfectly completed, but we think the current design already captured the most important information of YW. With this design, we are able to query any prospective and retrospective information that we are interested in.
The challenge now is how to map with YW model with ProvONE model, which is the mainstream model of provenance data. Our concern is an user without any knowledge about our YW model (vocabulary, structure, etc.) and only know about ProONE, how that user can query and get the information that he/she needs? Therefore, mapping YW model and ProvONE model is critical. So far, there are some YW model components able to map, but some of them are not. One of solution that we are looking at is using owl:sameAs to map YW & provONE vocabulary. However, we also realize a challenge that: even use owl:sameAs, SPARQL query doesn’t support a “transparent” way to retrieve “synonym” classes between the two models. We need to look at other solutions such as using a reasoner to derive additional truths about the concepts that we are modelling.
Besides these big tasks, we also ran some small experiments to explore rdfs:subClassOf behavior; using RDF and SPARQL for hamming numbers example. I would love to discuss more details about SPARQL and its query capabilities with anyone who interested in. So please feel free to leave any comments you might have.