Personalization for Dummies

futurelab default header

by: Gary Hayes

Ok your not a dummy but you may need a little primer in some of the basic personalization terms and mechanisms – these are broad enough to include Personal TV applications, eCommerce on the web or Broadband/IPTV and mobile video/audio and gaming.

Metadata, metatagging Without this personalization would be nigh on impossible. All content in the system will have a range of data about the data describing what it is, how suitability for certain users and an extensible range of other descriptive info. Users likewise can be described in a similar way but their profiles of course are constantly changing and being updated with new elements. MPEG7 is the next in the infamous sequence, a new standard that will be the container for media metadata

Rules-based engines: When a user has a crude profile already in the system data such as where they live, age, preferences can be applied to simple Boolean logic maps to target relevant content to them.

Collaborative filtering: An engine similar to ones found on Amazon and other larger eCommerce sites that aggregates customer information and makes recommendations based on the preferences of like-minded customers. This has evolved to where recommendation agents, other like-minded people, are used as a first port of call by many users.

Cookies: We all know what these are. A small file residing on the media server (local or remote) providing a way to track and capture a users path through content. The raw data in these cookies can then be used (with user consent) to provide a more relevant personalized experience.

Clustering: OK how predictable are we all. Sadly, very. Clustering is a way of grouping users together based on what they do. A more crude form of personalization than individual tracking and feedback it results in the ‘people who liked/bought that, liked/bought this’ type of resonance. In a world of mass niche this is one of the ways to group together interest groups by behaviour and likes/dislikes.

Real-time personalization: Where it get’s spooky. Based on what users are doing the interface or content presented changes dynamically in real time. Very resonant and where future interfaces and story-telling type applications are heading. Of course the more dynamic the system the more it resembles artificial intelligence of course.

Original Post: http://www.personalizemedia.com/index.php/2005/09/11/personalization-for-dummies/