Last week I, as Semantic Desktop guy of the DFKI, was guest scientist at two institutes in Hannover, the L3S Learning Lab Lower Saxony and the Knowledge Based Systems Group of the University of Hannover.
L3S and KBS are connected by Wolfgang Nejdl, who is a central person in both. We met during the preperations for an european union project and he invited me to visit him, a chance we used to gather ideas for the future of the Semantic Desktop, and to write a paper or two.
I gave a talk at their l3s info lunch about the search capabilites and future of gnowsis.
For interested readers I can publish the presentation as PDF, write me if you need it. (beware: hard spam filter, be precise).
Basically, Nejdl’s group is following a similar approach we do in the EPOS project and previously in FRODO and the weak workflows. Connecting desktop resources using semantic web technologies. They focus on the aspects of searching and information retrieval, whereas DKFI much focuses on non-obtrusive user support, personal ontologies and knowledge matching. Prof. Andreas Dengel‘s ideas about our subjective view of the world, what I called “personal paradigm”/”personal philosophy” at the beginning of my thesis work.
With the gnowsis platform we laid the ground for many possible experiments and since we are open source now, we could use some of the DFKI results also in the L3S context and vice versa.
My main contacts at L3S are Prof. Wolfgang Nejdl and Paul-Alexandru Chirita, both very interesting people and I am thankful for the last days with them. Prof. Nejdl was occupied with many things, like all Professors, but reserved some time to lunch with us and gave me some insight into his ideas and current work at his group.
Most of the time at Hannover I spent with Paul-Alexandru discussing an idea for a simple desktop search engine, based on gnowsis philosophy on desktop search but with a much simpler architecture than the complete gnowsis installation.
The existing adapter of the gnowsis framework allow the most generic interfaces to extract information. But the generity, as also discussed in the ESW wiki, is far too generic as to be performant and easy to implement. If you implement all aspects of an adapter and extractor you have a nice tool but also some work. Btw: Sven Schwarz and me documented this in a paper submitted to ISWC 2005, I hope it gets accepted so that I can say more about our practical experience there. In the last weeks I came up with a much simpler approach that I have to try out in the next months.
As we at DFKI are always interested in a desktop search engine and many projects of us are in need of one, we will probably start an open source project soon that will do the basis. To tell the truth, we already started writing the outline and basic classes.
My visit to Hannover brought some more things: photos, impressions, thoughts about the world etc.
Hannover has a nice city center with “old buildings”, pre-war buildings, and I walked some romantique streets with cafés and shops. If you ever visit Hannover, go the red line trip in their city center. For us tourists, a tour is painted on the street with a bright red line, you just have to follow it. The city map is synchronised and you can read descriptions about the buildings while walking. Very user friendly. I never had the chance to see Hannover like this during our CeBit visits.
photos soon to come on flickr.
