PowerSet Poised to Change Internet Searches

by , 3:00 PM EDT, July 1st, 2008

Powerset.com has been quietly building on its natural language and semantic search engine. Launched formally on May 12 and having added an optimized iPhone interface, Powerset adds the power of semantic indexing to Wikipedia now and more to come in the future. While that sounds fairly technical, the power and simplicity of Powerset comes from actually using it.

TMO spoke with Dr. Bruce Horn, Manager, Semantic Platform, Scott Prevost, General Manager and Director of Product and Mark Johnson, Product Manager of Powerset to find out more. If the name Bruce Horn sounds familiar, it should. Mr. Horn was one of the Apple engineers who worked on the original Macintosh Finder back in the early 1980s. Now he's the manager of Powerset's Semantic Search Platform.

Traditional vs Semantic Search

We're all familiar with the power of Internet search engines like Google and Yahoo! Typically, that search produces a Web page, based on certain criteria, based on some key word inputs. For example, "When was Sean Connery born?" will produce a page at Wikipedia showing the result, August 25, 1930.

However, when it comes to searching within voluminous archives, putting content into context, finding relationships between facts, and putting the whole of the results into a usable user interface, tools have been lacking. That's where Powerset comes in.

The Origins

Powerset is derived from technology first developed at the Xerox Palo Alto Research Center, PARC. That's also where the mouse and the concept of the modern GUI, with the Xerox Star, was developed and later turned into a commercial success by Apple. After much preparation, co-founder and CTO Barney Pell along with co-founder and Director of Natural Language Technologies Gian Lorenzo Thione have brought that technology to the Internet.

For example, when the semantic engine is unleashed on the 2.5 million pages of Wikipedia, the Internet's third most popular information source, an enormous amount of connected information can be extracted. For example, a simple search on "Henry VIII" from the main Powerset Website reveals a page with nicely laid out relevant information and connected resources that extend well beyond what one might find on a standard Wikipedia page. That's because Wikipedia pages (and most Internet pages) tend to be narrative and linear instead of indexed information.


Accordingly, one of the key features of Powerset is the term "Factz," items that have a semantic relationship to main topic and which can be explored further. The resulting information is summarized in such a way that someone doing research on the main search item is naturally led to the related and relevant items that are worthy of further exploration.

For example, expanding the Factz on Henry VIII leads to the term "created" which leads to more details on the "Church of England."

The Future of Powerset

For now, Powerset is limited to Wikipedia. TMO asked Mr. Horn about the business model, since Powerset is currently free. "First we developed the engine," he said. "Then we'll focus on generating traffic. Eventually, we'll monetize with an ad supported site."

Asked about what's next, Mr. Prevost said, "We're not ready to talk about that right now. But we do have some exciting plans for the next data source we'll tackle. Stay tuned."

There are over a million Websites with billions of pages. Finding simple facts is easy, but Powerset, by using the power of a semantic search engine, promises to present the important elements of huge volumes of information into a format that's more digestible by and useful for the human mind.