Comments on Wiki derivatives session

The session: This session covered Wikipedia and other Wiki derivatives within the context of medical education.
The first presentation, given by Konstantina Stampouliems, presented Wikipedia and specifically its medical articles content.   The general Wikipedia philosophy and policy were covered, as well as criteria for uploading content and the concept of an evolving document through open editing. The speaker encouraged all to contribute medical content and/or edit existing content and stressed how easy it is to do so – just press the edit button. Of interest was the questioning following on from the presentation, which focused on the scientific quality and correctness of the medical content.  There was concern that Wikipedia takes no responsibility for the quality of content, has no control on authenticity of users, and provides no information to users on quality. One questioner wondered if the primary attraction of Wikipedia is simply that it is free.  It was noted that there is some excellent content and some viewed the openness for all to upload, edit and access as a major benefit, but concern was expressed regarding what the company was doing to improve the quality.

In the second presentation of the session, Soren presented DBpedia and Life Linked Data. He motivated the technology by discussing limitations of google, for example, for retrieving information from multiple sources and/or where contextual knowledge is required in order to interpret the query.  He also overviewed  developments from which DBpedia has emerged by charting the evolution of the Web from version 1.0 through to version 3.0. The DBpedia way of addressing the accessibility problem is to move towards a web of interconnected knowledge bases. It achieves this through the RDF triple data model and deploys content negotiation.  Soren identified  the Iife sciences domain as being very mature as a target domain for this approach. He also outlined the Star rating for open data deployed, the current effort to linked data sources, and the distribution of this effort among  interested parties, such as publisher, third party, general public.  Soren introduced recently launched results of this initiative, i.e., DBpedia Live which transforms Wikipedia to linked knowledge base and supports queries using the SPARQL query language.  Examples and cases were presented and Soren concluded by elaborating various related initiatives, such as faceted browsers, a DBpedia benchmark, the concept of Semantic Wikis and Semantic Data Wikis, and Wiki tools. Once again the issue of content quality was raised during the following questions.

In the third  presentation, Pablo discussed Semantic Enhancements with BDpedia Spotlight. The presentation explained what is meant by semantic enhancement  (SE) and  how DBpedia spotlight, which exploits SE works.  Pablo’s view was that SE is essentially a means of making information easier to understand. It does this by assisting people to determine what is meant by providing access to associations, definitions and background. He illustrated SE with an online newspaper example where key text is highlighted and linked to related information. This approach has been “bridged” to medical sources, allowing users to traverse relationships to go from one text to another related text. This was illustrated with an example which elaborated relationships between fish oil and disease and inferences that can then be made. Pablo next introduced DBpedia spotlight which deployes this approach, using data from DBpedia and associated text.   This approach and the way spotlight works was further illustrated using an excellent example in which relationships between information relating to Lennon, McCartney, New York and Apple Corp were explored. In particular it was interesting to see how ambiguities are resolved by going to Wikipedia. The session concluded with a hands-on exercise using Sztakipedia, a Spotlight tool bar for Wikipedia to provide DBpedia searches.
The session concluded with a short demonstration of one of the mEducator tools for searching properties of drugs through navigation between related information content.

Comments: The first presentation was valuable as a stimulus for discussing the key question of Wkipedia content quality. Thie second presentation covered an impressive range of initiatives and developments. Perhaps more that the listener can digest in one go? Also, I felt that clarity could have been improved by including concrete examples of querying DBpedia at an earlier stage in the presentation. However, a value of this presentation was as a stimulus to search out information about any of the many initiatives covered. The third presentation covered important developments with good informative and illustrative examples. This was a valuable and informative presentation. The final short demonstration was a tangible and impressive illustration of some of the advantages of linking data sources, but could have done with more time to introduce, motivate and contextualise the tool demonstrated. In general, it would have been useful for delagates to have access to abstracts and/or copies of slides for the presentations, e.g., online., rather than relying purely on the presenters' narratives during the presentations.