Yesterday, I mentioned how Cleary Gottlieb had borrowed knowledge engineering techniques from the military, to capture the expertise of senior staff, embed it in a computer system and pass it on to junior lawyers online, in the form intelligent online textbooks and knowledge maps. But I was left wondering about the system’s interactivity, intelligence, and the currency of information therein.
Today, having read about iLink – another military-funded project – I’m now wondering if the Cleary Gottlieb system incorporates any of the social-networking ‘AI’ features mentioned in that article (given the military connection!)?
iLink was developed as a part of the SRI-led CALO (Cognitive Agent that Learns and Organizes) programme and was funded and managed under DARPA’s PAL (Personalized Assistant that Learns) programme. Mark Rutherford reports that:
iLink is a machine learning-based system that models users and content in a social network and then points the user to relevant content, discussions, and other network members with shared interests and goals across a broad range of scenarios.
To do this, the system uses message-matching technologies for finding related information, and algorithms for gathering data from multiple sources and compiling it together, whilst differentiating private information from that which is safe to share.
Now certain of those technology features/capabilities don’t sound too removed from some social software tools currently on (or almost ready for) the market. For instance there’s:
- Newsgator ES – has smart feeds and recommendations.
- PagesPlus – allows content to be pushed out to the categories and pages corresponding to the tags, and to the users who are subscribing to feeds from those categories.
- IBM’s Beehive – currently in the research phase – with the capability to ‘recommend’ connections based on activities, tags, bookmarking, etc.
- Zemanta – blog posts, articles or web pages are directly “read” by Zemanta, which recognizes all contextual content. It then combs the web for the most relevant images, smart links, keywords and text, instantly serving these results to the user to enrich and inform their content.
Nevertheless, it is iLink’s learning capabilities, and SRI’s work in modeling how real-time, dynamic social networks communicate and cooperate to solve problems, that really spark the imagination. Sarah Perez indicates that the technology:
[has been] used to develop a system for FAQ generation within a network – they call this technology “FAQtory”. With this technology implemented on a social network, FAQs are continuously generated and revised by the community using a Wikipedia-like model, as opposed to being static creations made by the site’s authors. [But it's no ordinary user-generated FAQ system] – instead, iLink’s FAQtory technology allows for incremental bits of information [to be added] – even those that don’t qualify as answers to the question. As the members contribute these bits of information, the learning system in iLink monitors how users are [attempting] to resolve queries and is then capable of drafting off the social network’s learning.
Potential commercial benefits and applications of such learning technology abound in business. Like expertise identification, comprehensive client information aggregation and delivery, FAQ generation and smart RSS filtering. As members of, and information in, social networks increase exponentially, there is a growing need to move away from search and retrieval models of information and expertise location. This is where smarter social technologies will help to streamline the process of recommending, and delivering, information and expertise (as well as filling-in information gaps as they go) to help people get their jobs done more effectively and efficiently.