Microsoft looking into just-in-time information retrieval

Last night I finally got around to watching Microsoft’s Comdex presentation, specifically the section where Susan Dumais shows off her new search technology “Stuff I’ve Seen.” (Search for “switch gears” at the bottom of the transcript or go to 1:07:50 on the video.)

Most of Stuff I’ve Seen is concentrating on the problem of quickly indexing and searching your entire hard drive, regardless of media format. (I sometimes jokingly refer to projects like this as YAPIM, or Yet Another Project Invoking Memex, my own thesis work fitting that description as well.) However, the part that interests me most is what they’re calling implicit query. As CNET describes the Comdex demo:

In demonstrating Implicit Query, Dumais began to type an e-mail asking a colleague about a set of slides for an upcoming conference. Before the message was complete, the program — which appears in a window on the side of the screen — pulled up e-mails, slide decks and Word documents containing the name of the conference and the future recipient. Each hit came with a brief summary of the internal content, date, the type of software the file was written in, and its potential relevance, among other information.

This is the same functionality that in my PhD I call Just-In-Time Information Retrieval, and is the main focus of the Remembrance Agent software I developed. It can be incredibly powerful (I use it regularly to suggest email discussions related to my blog entries, for example) and I hope Dumais pursues it. It looks like she’s still in early stages with the concept though, and and more importantly the current interface is still designed for explicit query — far too intrusive for something that runs all the time in the background. By contrast, Autonomy has had an actual product in this area for over three years, though I’d say the interface is still the real trickiness for this kind of application. Still, as is often the case one of the more interesting aspects of Microsoft doing something is that it’s Microsoft doing it. If implicit query makes it into a future version of the OS (and if MS doesn’t screw it up they way they did with that annoying paperclip) that’ll be quite interesting.