I’m still reeling from the ideas and insights that marked the hugely successful RecSys’07, a conference on the state of recommendation engine tools and techniques, which attracted a select crowd of 120+ participants from 16 countries across five continents (and over half represented industry, making the conference a perfect fusion of theory and practice). The air was electric, the energy was charged, and attendees eagerly compared it to the atmosphere before search – and Google – broke on the scene a decade ago to leave an indelible mark on industry and society at all levels. I was privileged to witness this world-first incubator of recommender thought, and will dedicate this week to its coverage.

I have tons of notes and interviews to sift through, so let me begin with a brief overview of a few of the ideas and issues that took center stage during the event. [I'm pleased to report that several participants from the likes of Amazon and MyStrands have also agreed to guest blog for this site, so we can all look forward to a steady stream of good ideas (some quite provocative) over the next weeks.]

Speaking of MyStands, who were kind enough to invite me to the event in the first place, here’s a quick update on progress there. The company is tight-lipped, but I expect it’s the quiet before the storm. MyStrands has hired some sharp minds from AOL and Nokia, building a team of over 75 people. The sole focus of their efforts: an agnostic and cross-platform recommender and an increased consumer push. Let’s just say ‘watch this space’…. (My thanks again to Gabriel Aldamiz-echevarria, a MyStrands VP, for filling me in and promising to circle back with the “scoop” when the time is right.)

In the meantime, here’s a radical thought courtesy of Tim Vogel, Chief Scientist, Aggregate Knowledge: Search alone is just a navigational tool – no matter the platform that delivers it. The winning model balances search and recommendation – all the better if the recommender calls the shots. But forget collaborative filtering as we know it from Amazon (and have since retrofitted for mobile).. Tim reckons the algorithm for success (no pun intended) is a new approach: People who searched like you bought that. If actions do indeed speak louder than words, then why not track them to add value to content suggestions. Put another way, turn the trail of search queries into a feedback loop and suggest content to users based on their search patterns. On a mobile device, where real estate is limited and so is user patience with input mechanisms (tiny screen, tinier keypad), this approach could pay dividends.

And who said it has to be content as we know it? Both Tim, and Shail Patel, Platform Leader, Unilever Corporate Research, hinted that personal advice may be the content that steals the show and attracts the traffic. Users not only respect the advice of experts, they will need it as the dismal state of public healthcare forces them to take more charge of their physical and mental well-being. As Shail put it: Unilever is focused on enabling “personalized mass-scale offers.”

An example is a newly launched site and ambitious scheme to assist users to achieve lifestyle goals and manage their weight. It offers users personalized advice (via recommenders) and incorporates an array of devices (via Bluetooth) to monitor personal progress. It’s not a mobile-only offer yet, but mobile is definitely part of the Unilever vision for where these services are going – soon. (BTW: Another very cool concept involves a Pond Institute in Madrid where personalized advice and recommendations result in a customized skin care program and the real-time production of a skin cream made in the factory and mailed directly to the individual customer.)

Check back tomorrow for more highlights, including a look at AOL’s personalization strategy and another radical idea: Is there a real risk that recommendations will enable popular content to become more popular, and less popular content to disappear? Put another way, in our eagerness to expose the legendary long tail of content can we actually achieve the opposite outcome? Think it over – and let me know what you conclude.

This entry was posted on Monday, October 22nd, 2007 at 9:15 pm and is filed under Mobile Search, Personalization, Recommendation, Web 2.0. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.