Can Watson identify related content?
Join the DZone community and get the full member experience.Join For Free
IBM’s Watson has undoubtedly been one of the coolest technology developments of recent times, and I’m tremendously excited about the possibilities for its use in healthcare in particular.
One feature of Dr Watson is its ability to locate the latest and most relevant research in a particular field so that this knowledge can be rapidly brought to bare on a particular challenge.
It appears to be cutting its teeth on a somewhat less life changing task. You may have noticed at the bottom of each post here a selection of related links. These are articles that are in some way relevant to the current piece you’re reading, and the articles are typically chosen by a keyword association method.
The smart folks at IBM are working on a venture called Concept Insights, which they hope will provide a richer means of uncovering contextual links to other content.
It works by trying to discover the concepts contained within a piece of content, which it then places into a knowledge graph that it uses to link it up with others of a similar theme.
The Concept Insights project is one of a raft of new initiatives that have emerged through the Bluemix platform that is designed to enable rapid production of web and mobile applications.
It was born a few years ago when developers at IBM tossed around the idea of using information theory to improve the web experience.
This initially saw the team venture down the path of building a contextual social network, but as that isn’t really an area IBM wish to get into they quickly shifted towards providing better contextual understanding of online content.
The technology was tested on an online directory of some 3,000 scientists and engineers. Each person has a profile on the directory outlining their skills and expertise. The hope was that the tool would make it easier to identify those with complimentary skills for serendipitous collaboration.
The developers are now branching out into testing the technology with a range of news sites to see how it works with traditional content.
I think this is a fascinating field, as so much value can be derived from making connections, either between people or between ideas, that go far beyond simply web links on a news site.
This will be a project to watch with interest. You can check out more about it via the video below.
Opinions expressed by DZone contributors are their own.