SWiPE allows 'deep search' semantic queries using the Wikipedia UI

We use increasingly complex searches and we get frustrated if we can't get the answers we want from our initial query. Swipe may change all that by querying DBpedia through the Wikipedia UI.
Written by Eileen Brown, Contributor

If you struggle with RDF triples (Resource Description Framework) and SPARQL (Query language and protocol for RDF) do not despair. SWiPE (Searching WIkiPedia by Example) allows semantic and well-structured knowledge bases to be easily queried from within the pages of Wikipedia.

If you want to know which cities in Florida, founded in last century have more than 50 thousand people you will be able to enter the query conditions directly into the Infobox of a Wikipedia page. Swipe activates certain fields of Wikipedia that generate equivalent SPARQL queries executed on DBpedia.

Swipe is a middleware system that intercepts user's browser requests for Wikipedia pages to introduce an in-page SBE (Search By Example) user interface.  Swipe answers structured queries posed by the user through the SBE interface. It uses familiar Wikipedia pages to help you construct queries, returning the results to the familiar Wikipedia interface or a standalone web page.

Credit: Swipe

Watch the demo on YouTube showing how the query is executed on the Wikipedia entry for Robbie Williams.  Hovering over certain fields in the Infobox changes the text to yellow.

The Infobox field can then be edited to search for the constraints you require in the backend data store (DBpedia).  Note the 'un' in the 'Pop Rock' field.  In the example, the search looks for people that do not play Pop Rock.

Developers, Maurizio Atzori at the University of Cagliari, Italy, and Carlo Zaniolo at the University of California, Los Angeles reckon that Swipe seeks to 'maximize both ease of use and query power by letting users work directly on the Wikipedia pages displayed in their browsers'.

Three steps

They say that 'the user will specify the query by marking up the information-box shown in the page'.  There are three steps to the process:

  1. The user starts by loading an "example" page in the browser. This is a page that is representative of those that a user is looking for. Thus a user interested in cities might just retrieve from Wikipedia a page like "Boston," but any large city would do as well. This initial step provides the starting point for entering the actual query.
  2. The example page looks like the original Wikipedia page. However the Infobox of the page is now an active form that can be used to enter the search parameters defining the query.
  3. The user specifies the query by typing into selected value fields in the Infobox, and then issue(s) the query by a simple click. Swipe returns a description of the Wikipedia pages that satisfy the query conditions entered by the user.

The search prototype will be demoed at the World Wide Web conference in Lyon, France this month and it will be interesting to see how the demo is received. We use more and more complex searches in our web wanderings and we are frustrated when we can not get the answers from the initial typed query.

Moving us 'beyond the capabilities of web-search engines, their keyword oriented searches, and toward the much more powerful query capabilities of database management systems' is a positive step.  It brings us closer to web 3.0 which started us buzzing several years ago.

It brings intelligent, natural queries which are easy to construct and it brings search to massive backend datasets. Moreover, it brings information in these datasets to anyone with a browser -- and the right search plug in.

Related content:

Editorial standards