Miriam Fernandez, a PhD student from the Autonoma University of Madrid (UAM) and I (Vanessa Lopez, currently in my fifth year working at KMi), presented our paper at the main track of the ICSC conference in sunny Santa Clara, the heart of Silicon Valley (August 2008). This paper is the fruitful result of the two summers Miriam spent collaborating with us in Milton Keynes, plus a few nights without sleeping.
An enthusiastic and smily Miriam started to present our paper to a very attentive (maybe because of the lack of internet) audience. The paper describes a formal evaluation, adapted from TREC, for our novel semantic search system, which aims both to bridge the gap between the user and the Semantic Web and also between the Semantic Web and the Web itself. Our system integrates PowerAqua, our Natural Language interface to the open domain Semantic Web, developed in KMi, and the IR system developed at the UAM to support semantic search. The latter complements the answers from the Semantic Web with a list of ranked documents.
It was when Miriam stated that she believes that the Semantic Web can help to overcome the limitations of keyword-based search that I (inspired by the Nodalities article written by Mathieu about Watson-PowerAqua) decided to play devil’s advocate and jump out of the middle of the surprised audience "Excuse me but I have a question … I see the vision, but this is all geek things, how can this help the real user? How can this make a difference on the results I get when I query the web?"
Miriam’s recovered quickly from the shock and a slide with a flying pig on the sky, and a little devil asking a question to the happy pig appeared on the background. At this point, I could hear some laughs, and I realized we got what we wanted, an incredible attentive audience … Time to become serious … Miriam continued "what if I tell you that the SW can help systems to understand the real meaning behind user queries and provide the user with specific answers-as PowerAqua does when you ask the question "who are the members of the rock group Nirvana?"
We continued with a small discussion about the Semantic Web angels and daemons, why current semantic search systems are not currently competitive enough against traditional IR search engines, and how can the SW can be used to exploit and re-use all the available ontologies at a time and in real time? And finally, how can one extract meaning from Web content without converting the whole Web into a Knowledge Base?
Then, we introduced our novel semantic search system, describing its main features and illustrating how it can be evaluated against standard keyword-based approaches to search on the Web.
In conclusion: "two better than one" (even for conference presentations). Although there is still a lot to do to achieve a full integration of our two systems, we have proven the feasibility of applying background knowledge drawn from the Semantic Web to improve searching on the Web what happens when the Semantic Web meets the Web is as promising as Sally meeting Harry.
The presentation finished with quite a few interesting questions and even some original congratulations from one guy in the audience: "Nirvana does exist … in our heart", and the SW knows about it, how cool is that, right?
Related Links:
