Google working on semantic search

Google’s believed to be on the point of transforming the way it delivers search results, taking a leaf out of rivals’ books and providing answers rather than links.

According to the Wall Street Journal, it plans to include semantic search, a la Wolfram Alpha. The feature will sit alongside existing search algorithms, with answers appearing above the standard search results.

Search executive Amit Singhal told the WSJ that, over the last two years, the company’s been quietly setting up a massive database containing hundreds of millions of ‘entities’ – people, places and things.

Other semantic search engines, such as Wolfram Alpha and MC Hammer’s WireDoo, have failed to take off in any big way – although Microsoft’s Bing has been a bit more successful.

But they’re not the only competition, and Google’s move is believed to be an attempt to head off sites such as Facebook, which has its own massive database of entities and is becoming increasingly useful as a search tool, particularly when searching for people.

The new feature could even be seen as a way of competing with Apple’s Siri digital assistant, with Singhal telling the WSJ: “When we can deliver small nuggets of information, that system is far more suited to mobile phones and searching with voice.”

Google’s been thinking along these lines for quite some time. As far back as 2008, it bought Powerset, introducing some of its semantic search features the following year.

While further changes are expected to appear over the next few months, a full switch to semantic search isn’t expected for some years.