Describing the Descriptors (this is a copy of a Sep 2005 post – as it is almost FAQ in nature)
I have had a few separate emails asking exactly what I mean by personalization. I believe the area is broad (deliberately so – I find it hard to be constrained by narrow topics), portentous and covers the cross-over between life, media and virtuality. I spent three years as Chair of the Business Models working group with an international team developing a really cool standard TV-Anytime that became a way to personalize content and map content to viewers – but went beyond just TV or Radio very quickly. A tired old presentation at NAB 2003 – I had had a hard night in Vegas, as you do – still the last slide part begins the journey.
In lay mans terms though personalization means to me:
1 – the filters through which we have services and media tailored to us and
2 – the design of services and content that are flexible enough to resonate with others on a one to one, or one to few level.
Ummm – not really lay mans descriptors, how about:
“automatically getting stuff (out of too much) that you will likely want or need and how to make stuff that will reach and resonate with others through the same mechanism – agents of choice use your preferences to point out stuff you will like”
This is way beyond search – not a world hoping that Google or other search engines will pick up all those lovely keywords or have crawled your latest video or audio track – then having pulled them out of the broadband’osphere, display them for all to see – no this is… you can see I am struggling here! not totally happy with these descriptors – how about a structured little summary of the key alternate perspectives on personalizing media:
Personal journeys – services are responsive and tailored to you as individual when you are:
– On narrative paths across devices or through media types
– On routes through one media type (the ‘you’ cut) and the non-linear narrative variations around that
– Immersed in interactive services
Your profile – creating the digital you, the thing that represents you as entity and interoperable dynamic soul, extended into the digital space, enabling:
– Control – over this digital personality you decide who you are, or not, or if you want and trust the system to ‘learn’ you most effectively
– Interface – whatever you are interacting with, car controls, a games interface, broadband TV portal, ePaper etc: it adapts for you or is set for you to use it most efficiently
– Changing profiles – growing and changing as you travel across interoperable cross-media systems such as taking the ‘mini me digital you’ from home to mobile to work to social, learning about you as it travels
– Tailored services – that alter themselves based on your characteristics and carry on learning about you in the service
– Media match – having audio, video, books, interactive content or other merchanidise linked to you
– People match – having other profiles, people matched to you for collaborative creative, project or life goals
Content creation and customization – media that you create and share that reaches others – AV, text, image, re-edits, mashes, sampling, blogs, vblogs, wikis all richly meta-tagged to allow personalization engines to draw your media and match it.
“The media altered to fit your match users stated needs as well as needs perceived by the system based on the available user information, for the purpose of better serving the user by anticipating needs, making the interaction efficient and satisfying for both parties and building a relationship that encourages the user to return for more.”
Life digitised, non-visual proxies of you in the virtual space, a quest and goal of merging all those proxies into the most effective representation that creates the least resistance between analogue and digital. Some terms:
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.
© Gary Hayes 2005