The Semantic Web is currently the most evolving area of the Internet development. This second-Generation Web is based not on HTML but on XML and related technologies, such as RDF, Topic Maps, Ontologies, Namespaces, and others. The 'Semantic Web' is "a vision: the idea of data on the Web defined and linked in a way that it can be used by machines for automation, integration and reuse". The Semantic Web not only requires adapting existing visualisation techniques and developing new approaches, but also opens wide opportunities for their practical implementation. [1]
In this project, I would like to create a tool that starts with a file, spiders, builds and displays a graph of instance data. Once built, users will have the option to hilight certain patterns and relatinoships withing the graph. For exapmle, in a food web network, users could ask for visual indications of things like
Users could also look for patterns between related subgraphs, i.e. show me all of the fish who are herbivores (explicitly eat non-animals); then, find any species that consume only these types of fish. Essentially, find some interesting subgraphs and then find some middle node that connects them.
Using technology from RIC, users will be presented with a list of classes. Thy can shoose a class, then optionally choose a property (either with a domain or range of that class), and again optionally choose another class to be at the opposite end of that property. Users can choose several of these and see the patterns emerge in the graph. Thus, the graph is built of instance data, but is analyzed through the ontology. Inheritance will be preserved in this search, so one could search for "Animals that eat Plants" and see all of the fish that eat plankton.
More compelling examples emerge in large networks of information. A graph of the RDF from a university department could include a lot of information, including research projects, labs, classes, etc. Isolating the graduate students in the graph and highlighting them would itself be useful. Finding the links between these students and, say, research labs or project would show links that may be useful for collaborations, highlighting common interests, etc.
Once this is implemented manually, I plan to automate the discovery of these these bridges between objects to identify potential links between subejcts or concepts that the user may not know about or anticipate.
A tiny group. Jen.
Thanks to Dave Beckett for the suggestion of this topic and the reference to the symposium.