Digg.com is to launch its new recommendation engine later this week. It has a number of features that will help promote relevancy and diversity of stories to users. ‘Diggers like you’ lets you track stories 'dugg' by people with a similar viewing and voting history and recommendations are computed within topic areas to ensure that only your shared interests form the basis of suggested content that you might enjoy. Techcrunch has done a nice piece on this with some background info on the nuts and bolts of how the engine works http://www.techcrunch.com/2008/06/30/update-digg-recommendation-engine-confirmed-for-this-week/

Relevancy and personalisation are hot topics for sites with a high volume of content and a large base of users with wide ranging interests. They serve to help users filter the sheer scale of information and to increase the value of the content for each user. This is nothing new – Amazon’s ‘people who bought the also bought’ works to help users navigate the plethora of media on offer and Netflix have a $1m open challenge to the web community to improve the algorithms that underpin their recommendation engine by 10% in a bid to enhance the accuracy of suggestions for future viewing http://www.netflixprize.com/.

From our perspective, social media environments are fascinating opportunities for developing tools that tap the power behind the users collective behaviour – ratings, rankings, reviews, tags, bookmarking to deliver smarter ways to navigate content and to improve the user’s experience with every interaction. It’s not a simple task but one that will refine our understanding of how to create interesting applications from multiple data sets, how recognising patterns of behaviour can inspire delivery of new services and the ablity to provide different perspectives on the huge volume of data that users create over time

A book that’s been useful to us in our work in relevancy and recommendation is a book by Toby Segaran called programming Collective Intelligence http://www.amazon.co.uk/Programming-Collective-Intelligence-Building-Applications/dp/0596529325/ref=sr_1_1?ie=UTF8&s=books&qid=1214993611&sr=8-1

which achieves that rare task of communicating sometimes complex ideas in an easy to grasp way. Happy reading.

Posted  2/07/08