Week Five Update

Week Five Goals

The main goal for this week was to work toward integrating the DataONE data that I have extracted to work better with the various browsers I have found. In particular this was going to require that I understand better the vocabularies used by the RDF browsers and align the DataONE data to it.

Results

Using the Tabulator RDF browser, I began to see some results I was expecting. In particular where I could see locations on a map and dates on a calendar. The biggest benefit from this is that now I am able to show integration points between the different datasets, regardless of their source repository. Although what I can show is limited to points on a map and items in a calendar, it still demonstrates the qualities of RDF browsers accessing datasets from the Semantic Web. Specifically, pulling in these multiple data sets and plotting their longitude and latitude, I am able to see the graphical perspective of them all.

I regenerated all Dryad and KNB RDF data and placed them in the LOD4DataONE project on my server. I was able to add more links to dbpedia data and leverage browser knowledge based on vocabulary used. ORNL DAAC data will be extracted next, it was just difficult to iron out details of each repository’s api while moving forward with LOD integration.

As a result of this I have a more complete use case that will exhibit loading various data sets and plotting their data, showing the integration that is possible on the Linked Open Data Cloud. This process is quite involved and takes quite a bit of documentation. I will be updating notes, the use case and the GitHub source with all the details. I will also be building a demo that works through the use case in an effort to explain the work for this research effort so far. A blog will be posted when all documentation is complete, pointing you to different items.

Observations

1. KNB data, through the use of EML, is quite a large metadata framework. For this integration, I grabbed all data through the Metacat API and generated RDF for what I felt was useful for showing browser integration and data integration. Because all the data is structured in XML it was not a difficult task to extract the data and map it to RDF.
2. In working with different browsers, four to be exact, it is clear to me that work is needed to consider relevant views of DataONE data. Many of the browsers provide environments for viewing specific vocabularies, e.g., SpatialThing, even a Music Ontology, but not the sciences. This is not far from what it appears the RDF browser makers were expecting. In Zitgist, for example, templates can be added, in the Tabulator, a Javascript tool can be added to a tab.
3. It occurred to me, as I was searching for terms and relationships between data to create links, that tools that enable users to publish linked data should seriously consider a recommendation system to help users choose relevant links. Manually, I would search for relevant terms concerning a subject, then look for it in either Freebase or DBPedia. Although I did not find either tool easy to search through, I would imagine that they should have some type of automatic search capabilities that could be leveraged