By Luis Lastras
One of the most common features of online news sites, blogs and many other Web pages is the list of “related” items. Those links are produced using simple keyword associations, and, often, they’re basically the same information in a different package.
But what if you had a feature that helped you find items that lead you off in surprising new directions?
That’s becoming possible now thanks to a new service, Concept Insights, that my research team at Watson Group created.
Rather than using keywords to discover relationships, the technology digs deeply to understand the concepts contained in documents, maps those concepts in a giant knowledge graph, and surfaces linkages between the documents on Web pages. We think this new capability will be revolutionary–greatly enriching the online experiences of consumers, business people, scientists and students.
Concept Insights is one of five new cognitive computing services available today in beta versions on the IBM Watson Developer Cloud through Bluemix, our platform that helps developers of all types rapidly build, manage and run Web and mobile applications. The other new services include speech to text, text to speech, visual recognition and tradeoff analytics (which helps you evaluate alternative approaches to problem solving). These services join eight we made available to developers last October when we opened the doors to the Watson Developer Could.
IBM has made available an ever-expanding suite of Watson applications and services addressing broad swaths of the business world. But we understand that there will be countless other uses for Watson’s cognitive technologies that we won’t address directly. So we created the Watson Developer Cloud to make the core technologies available to others to build upon. The idea is to let thousands of cognitive apps bloom.
We’re encouraging developers from established businesses and startups, in addition to university students, to build apps in our Bluemix sandbox. So far, more than 5,000 developers (including IBMers) have taken us up on the offer.
Concept Insights got its start about two years ago when I started brainstorming with team members Livio Soares and Michele Franceschini about how to use the scientific field of information theory to enrich the Web experience. We were frustrated with the limitations of traditional Web search and networking tools. Our idea was to create technology for a new sort of social network–one where people discover each other based on their interests and expertise rather than their existing friendships.
Initially, we gave the technology the code name Piazza, because we thought of these social networks as public places on the Internet where people who don’t know each other might meet and interact–much like the central squares in Italian villages. We conceived of it as a town square for the world.
IBM isn’t in the business of building and running consumer social networks, so, as our technology developed, we shifted focus to providing core capabilities that could be used by businesses and consumer-facing Web sites.
To test the technology, we put it to use on a Web site that contains a directory of the 3,000 scientists and engineers at IBM Research’s 12 labs in 10 countries. The directory includes detailed descriptions of their expertise and research. The Concept Insights tool makes it easier for researchers to identify and others in the organization who have complimentary expertise to theirs, so they can collaborate in uncharted areas.
My team is testing the technology with a major news Web site and a scientific organization.
Concept Insights and the other services on IBM Watson Developer Cloud are examples of amazing new cognitive capabilities that are coming along. I believe that they will help transform the way we interact with each other and with information in the new era of computing.