UPDATE: DBpedia Search is currently offline. An updated version will be available soon at http://search.dbpedia.org.
Building the Semantic Web is about integrating data, that is already out there.
A large body of information and knowledge is often already available in structured form, yet not accessible as such on the Web. Wikipedia is one of the most prominent examples.
Integrating open data provides real value. It saves the time and effort to re-enter data that is already out there and it leaves the data and editing where it belongs: at its origin.
The DBpedia.org community project approaches that problem by extracting structured information from Wikipedia and making it available on the Web using Semantic Web technologies.
Being part of Linking Open Data project, DBpedia can and already does function as a link hub between different datasets.
Within the Semantic Web Community, DBpedia is already perceived as a useful source of data with powerful query possibilities (see Michael K. Bergmann’s excellent article about DBpedia).
However, DBpedia lacks a simple way to explore that data, to make it usable and useful to everyone, not only to developers. My primary object was to solve this problem when I started the development of a new Linked Data browser in early April. Thereby bringing together the structured data approach with the usability paradigm of the current Web 2.0 movement.
Furthermore, since Semantic Web has received heightened attention from the internet community and specialized media, start-ups and entrepreneurs have entered the scene, but they tend to work behind closed doors, to hide their ideas behind closed beta sign-in-screens and call that “stealth mode”.
Hence, to contrast that, here is the public, sign-in-free prototype release:
Search DBpedia.org
“Search DBpedia.org” is a web application that combines the strengths of structured and linked open data with a new data browsing experience. It features full text search capability, offers faceted browsing (not yet, but that’s to come), uses entity relations, shows result classifications, and allows to explore the DBpedia dataset and its linked data. It uses open Semantic Web standards like RDF, SPARQL and FRESNEL.
To give a short example: Search for “Scorsese” to find the director “Martin Scorsese” as well as films that are related to him. The result item “The Departed” shows that it is a film directed by Martin Scorsese. One could now select from a tag cloud showing all types of results weighted by their estimated relevance to narrow the set of results. So if one is looking for films related to Scorsese, simply select “film” from the tag cloud. One could also use “Martin” as a search term to find the director Scorsese or the film “The Departed”. Try that in Wikipedia’s search…
Another example: Looking for “Tim Bearners-Lee”, one get a details view of his data and also of data about his publications that comes from the linked dataset of DBLP.
Being a first prototype, there are some open issues, performance limitations and certainly many bugs. But talking about technology is so much better when having something real to talk about. Small but real things can start something big.
Like Mike Linksvayer said after the original release of DBpedia back in January:
The Semantic Web is so here, now. Doubters repent!
[...] Once people get hooked on access to a semantic encyclopedia they will want similar access to the entire web!
This article is intended to be the first of a series of articles describing the new user experience provided by “Search DBpedia.org.”
Some words about my background: I’m a research assistant at the Free University of Berlin, working together with Chris Bizer and Richard Cyganiak. I’m a founding member of the DBpedia.org project and the development leader in the areas of data integration and visualization.
I got a ‘502 bad gateway’ message on the first view of http://dbpedia.org/search/
A reload fixed it, and told me:
So this seems to be a known problem. You might want to change your links to this address for now as people might be scared away…
Thanx for the link, anyway
Georgi,
This is wonderful work for demonstrating the strengths of using semantic web technologies to organize and manage data. The tag cloud (and coming facets) are especially useful and important for the function of an encyclopedia as a starting place for people unfamiliar with a body of knowledge to get an orientation to it. I tried to imagine myself as one of my former students in that role, and came up with a neat example of this tool being more valuable than a Wikipedia search that seems to be better, but ultimately is weaker.
I used to teach Anglo-Saxon literature, so I did a search for “Cynewulf.” Wikipedia’s search took me right to the page about the poet, which included a link to disambiguation pages. Seems good at first, but if a student is just being introduced to Anglo-Saxon studies, they miss awareness of other Cynewulfs.
Searching DBpedia, on the other hand, also offered the tags “bishop,” “monk,” “poet,” “saint,” and “sovereign” for filtering. From a professor’s perspective, I think that’s a much better introduction because it would make a student aware of the different Cynewulfs right away (I have little confidence that a student would be so diligent as to follow Wikipedia’s disambiguation link). Hopefully, then, they’d be less likely to be confused by ambiguous references when they go on to doing research in the scholarly journals. At a glance, they’d have learned much more about the field of Anglo-Saxon studies.
It’s that expression of relationships, rather than data in isolation, that makes me most excited about these tools…it will bring big changes to how university faculty and libraries work with students (at least it will if they pick up the ball and run with it!)
Thanks much,
Patrick
Thanks a lot for this great feature.
But I don’t understand one think :
Why does the request http://www.dbpedia.org/search/search.aspx?QueryString=Pete%20Postlethwaite&QueryClass=actor returns firstly movies, even if I precise QueryClass=actor ?
Hi Ranfaroth,
well it seems you spotted a bug, which I unfortunately can’t reproduce. When I search for ‘Pete Postlethwaite’ and select ‘actor’, only one result is returned. So if you could tell me how you got there it would help a lot.
Cheers,
Georgi
I just do the same manipulation, which brings me on the URL I gave above, and I still have many results, with movies first, as shown on this screen : http://img514.imageshack.us/img514/136/dbpediazu7.png
Search seems broken…
Hi Nolan,
yes, unfortunately it is broken at the moment.
I hope to bring the search up again soon.
Thanks,
Georgi