These days there area great many events to choose from if you want to find out about the latest innovations in search and information access. But what makes Search Solutions different is that it provides an informal, interactive alternative to the highly choreographed, glossy brochure, convention hall style of conference that seems to dominate the search industry. Search Solutions is a much smaller, more intimate event, structured so that it is possible to meet just about all the speakers and delegates during the course of the day. In addition, the focus is firmly on the presentation of original material with credible technical content, rather than ‘sales pitches’ that contain no element of self-critique or realistic objective evaluation.
In common with previous years, this year’s event attracted an eclectic mix of speakers drawn from global organisations, start ups and solo players in the world of web & enterprise search. We were privileged to have BCS President Alan Pollard formally open the proceedings with an overview of the Society’s ambitious plans to reposition itself as the “Chartered Institute for IT”, providing an overarching certification body for all the specialist professions within the field. The technical programme was kicked off by Mike Taylor, who presented a overview of some of the techniques that have been used for learning ranking functions at Microsoft Research. It had been a while since I’d seen anyone provide serious mathematical analysis in a ppt presentation, but Mike more than made up for this, giving us enough equations in 20 mins to last a fair few conferences to come.
This was followed by Vivian Lin Dufour, who presented details of Yahoo’s progress in developing an interactive search assistant known as Search Pad. The idea of this app is to detect when users are engaging in research activity and support them with automated tracking, authoring and annotation tools. I think it’s a great idea, although not entirely unprecedented (the idea of automatically detecting and reacting to human behaviour has been a popular theme of AI and HCI research for many years). To me the hard part is reliably detecting the patterns of behaviour and defining what those patterns mean in the first place, as well as finding an appropriate balance between user control and system initiative (which will inevitably vary by user, task and context). Rounding off the opening session was Richard Russell from Google, who presented a highly entertaining overview of how Google ads work and the ecosystem that supports them.
The afternoon session was opened by Greg Grefenstette of Exalead, who presented his thoughts on the intersection of database and search engine technology, and how the two can be combined to create “search based applications”. These provide a unified view of structured and unstructured data, updated in real time and accessible via a familiar search UI (a value proposition not entirely dissimilar to Endeca). One key difference however would appear to be in the emphasis on user experience: the Exalead position seems to be that UX is more of a ‘hygiene factor’, indigenous to the platform itself, and that end users will simply configure the UI however they see fit. But I find it hard to believe that this approach would scale for all but the simplest of applications –I have seen too many examples of poor usability to believe that good design will happen as a fortuitous by-product of end user configuration. Of course, this may all be a matter of interpretation – if you define the scope of UX as starting with query formulation and ending with results presentation, then fair enough; it probably is relatively trivial. But if you define UX that way then you probably have also missed the point.
One of the most memorable talks in the afternoon session was from Richard Boulton, who described his experiences in developing a web service interface to IR code library Xapian. This could have been a relatively dry topic, but was brought it to life by Richard’s personal anecdotes of the eccentric away days and bootcamps he attended in developing the system. The proceedings ended with a panel session on Interactive IR, picking up on the threads that had been generated on the IR mailing list around the subject of the “Google Generation” and the how the information seekers of the future deserve something more than simply a further extension to the best-first ranked results paradigm. The discussion that ensued extended far beyond the original theme, with many delegates joining in to offer their views of how the world’s information might be organised in the future.
The day closed with a drinks reception in the atrium and the IRSG AGM, at which I’m pleased to say many new delegates came forward to serve on the IRSG committee for the forthcoming year. If you’re similarly inspired and interested in joining them, drop me a line. We’re particularly interested in building stronger links with the IR practitioner community. Next year’s Search Solutions will again be held at BCS HQ in late October (date TBC). We’ll be starting the formal planning early in 2010 so if you’d like to be involved just drop me a line either here or at irsg@bcs DOT org. In the meantime, we are hoping to run an Industry Day at ECIR 2010, in Milton Keynes on April 1st. Watch this space for updates!