I was checking out a research paper in the ACM Digital Library: SibylFS: formal specification and oracle-based testing for POSIX and real-world file systems and spotted something interesting: a “Concepts in this article” drop-down “Powered by IBM Watson” For more on Watson, see IBM Watson: Beyond Jeopardy! Q&A
I’ve been thinking what calls to action a user in our audience might be interested in when they land on a page for a research manuscript in our repository.
One thing they might want is show me what this is about or find other articles similar to this paper following a thread of interest. A semi-automated way of doing this would be to offer some concepts to choose from. This can shade into data mining, such as OpenMinTeD.
Other automated clustering and keyword / concept generation techniques exist. I’ve only ever seen one attached to a repository (Elsevier Fingerprint Engine), I’ve seem demonstrations of Pingar and Clade (2012 blog) and I know of some interesting recent commercial work in the area.
I wonder why we don’t see concepts pulled out automatically from repositories and offered to users for navigation. Is it too difficult, too expensive or just not helpful to users?