Would you like to ask Wikipedia tougher questions than today's simple keyword searches allow? A prototype plug-in that can do just that will be demonstrated at the World Wide Web conference in Lyon, France, next month.
Called Swipe - loosely short for "searching Wikipedia by example" - the software aims to let users of the online encyclopedia answer complex questions that most search engines would stumble over. For example, trying to figure out "which actresses won academy awards when they were under 30 years old in the last 25 years?" becomes relatively simple when using the program.
To use Swipe, questions are not typed out in the form of the natural language above, but Swipe is nevertheless designed for everyday users: no knowledge of arcane database query languages is necessary, say the developers, Maurizio Atzori at the University of Cagliari, Sardinia, and Carlo Zaniolo at the University of California, Los Angeles.
The pair wrote Swipe using MediaWiki, the software Wikipedia is based on, but it draws its answers from DBpedia, an expansive collection of 3.6 million data entries harvested from Wikipedia's pages. The data pop up in the info boxes on the right-hand side of Wikipedia entries, which list the details we use to describe the world, such as dates, prices, ages, heights, names, places, distances, bit rates, bytes, running times and geographical coordinates.
In a demo video, the pair show how to use pop singer Robbie Williams's Wikipedia page to find a list of "musicians" with the first name "Mike" who don't play "poprock" - by overwriting his name and occupation, and placing the construct "un" before the genre "poprock" (watch the video above).
Swipe could easily be made available as an option on Wikipedia one day, says Atzori.
"I like the idea because of its simplicity and the way it uses familiar Wikipedia info boxes to construct powerful queries," says Nigel Shadbolt, head of the web and internet science group at the University of Southampton, UK.
"Structured data resources like DBpedia represent a really rich knowledge base," he adds, but they are difficult for the average internet user to take advantage of. "If we could build a query engine that's usable by all, that would be a real winner," he says.