|Magpie from Open University|
What's the Problem?
Browsing the web involves two main tasks: finding the right web page and then making sense of its content. A significant amount of research has gone into supporting the task of finding web resources through 'standard' information retrieval mechanisms, or semantics-enhanced search. Much less attention has been paid to the second problem. Magpie, is a tool which supports the interpretation of web pages. Magpie acts as a complementary knowledge source, which a reader can call upon to quickly gain access to available background knowledge relevant to a web resource. Magpie works by automatically associating an ontology-based semantic layer to web resources, allowing relevant services to be directly invoked within a standard web browser.
Towards a Solution
Magpie allows users to browse the web through an ontology. Figures 1a-c below show screen snapshots of Enrico Motta's home page viewed through the AKT reference ontology (selected earlier). In figure 1a the concepts 'People', 'Organisations', 'Projects' and 'Research Areas' have been highlighted using colours to reflect the type of item recognised. In figure 1b, the 'ScholOnto' project has been selected and a menu of available semantic services is displayed. Finally, figure 1c shows the results of selecting the service 'Shares Research Areas With'. The new window shows all projects which have overlapping research areas with the ScholOnto project. The projects are ordered according to the number of shared areas.
Figure 1a, 1b, 1c. Showing Enrico Motta's home page viewed through Magpie. Click on a screen snapshot to show an enlarged version.
Figure 2a and 2b. Showing Enrico Motta's home page viewed through Magpie with Collectors. Click on a screen snapshot to show an enlarged version.
When Magpie is used every web page viewed passes through a proxy which extracts items matching entities within a populated ontology. This process enables us to create trigger services which initiate actions dependent on patterns of data extracted from a web page. Magpie Collectors automatically store extracted items which match a semantic filter. Figure 2 shows Enrico's home page viewed with three Magpie Collectors which store information on:
In figure 2a we can see that Enrico's home page mentions one person (Enrico Motta) and 6 projects. Enrico Motta has another 10 projects which do not appear on his home page. Figure 2b shows the result of selecting the link to Enrico's publications page. The user has selected the item 'Enrico Motta' within the first Magpie collector. In addition to the available semantic services the urls of where the item appeared is shown. We can see that the term 'Enrico Motta' appeared in both Enrico's home page and in his publications page. This functionality enables Magpie to provide a semantic bookmark service where any item previous browsed can be retrieved through a semantic query. For example one could ask "Could you find me the page I saw last week which described a researcher working in the area of knowledge modelling who's just published a paper in IEEE Intelligent Systems".
Take a Guided Tour
Coming real soon...
Try a Demonstration
Version 1.0 released January 2003. Internet Explorer v 5.0 or above.