Attention Landscape: Personal Relevancy for Alerting

Another in a series of articles I'm calling 'Attention Landscape'.

Applications of Personal Relevancy for Alerting
Tuning out the noise with Measured Disruption

With information reaching saturation point, time is the scarcest resource of our generation. The only way to better use our available time therefore, is to find ways to hyper-efficiently allocate our attention.

The best most pervasive example of hyper-efficient allocation of attention by a product or service has been Google. Google can determine your ‘current’ interests based on the keywords you type into their search box to deliver information matching your current focus of attention with uncanny accuracy. This type of attention might be called ‘Current Attention’.

The next challenge, however, is to allocate ones attention to information that may not be of current interest, but rather of ongoing or general interests to a user during their daily lives.

This type of attention might be called Ambient or Passive Attention.

Allocating Ambient Attention should be based on the ongoing interests of the user, the task their performing and the level of disruption the given piece of information deserves based on its personal relevancy (taking into account their level of interest and workload).

With this in mind, there are four technology challenges in the provision of Ambient Attention Management.

1. Personal recommendations
2. Task or activity based recommendations
3. Personal Relevancy
4. Measured disruption (Providing information in a way that reflects its immediate relevance)

Personal Recommendations
Information such as breaking news about world events that affect you, entertainment that interests you, people that are close to you and events that require your attention are all part of your ongoing interests.

In this case, what’s needed is a model of your previous articles of interest so that similar information can be crawled and recommend based on common themes and topics.

This type of approach has been built into services like Findory and Rojo and goes some way to collecting a broad range of content that is generally relevant to your interests and presenting it on the screen in a customized newspaper.

Task or activity based recommendations
While personal recommendations take into account what a user is generally interested in to find similar information that might be relevant, Task or Activity based recommendations consider what a user is currently doing to make recommendations for content that is relevant. Are they writing a document about Tree Frogs? Perhaps they would like to know all relevant research, statistics and news about Tree frogs?

This type of service would go some way to helping knowledge workers gather a context sensitive periphery view to their current work that might include previous resources from the corporate intranet, online research and statistics or even just images to pretty up their report.

Besides one or two low profile examples, this opportunity is largely untapped in the marketplace.

Personal Relevancy
While Personal Recommendations finds additional information that might be relevant to you personally. Task/Activity based recommendations finds information that is relevant to the task you are currently performing. Personal Relevancy is about determining the exact value of information to you at this very instant.

A Personal Relevancy algorithm should boil a number of factors down into a number that can be plotted along a fixed continuum from least relevant to most relevant. By plotting an item’s Personal Relevance against a know range, the user and/or a software agent can make intelligent decisions about the presentation of that item in order to maximize their time.

Measured Disruption
Explicitly subscribing to information you care about (E.g. RSS feeds), Personal Recommendations and Task/Activity based recommendations all converge to create a subset of the world that may interest you to one degree or another. The challenge remains, however, to find a way to consume this information in a way that allows you to maximize the time in the day. It would be simply impossible to keep track of all the information you care about if you were to stop and read every article about every topic related to your interests and activities.

Measured Disruption is about finding an approach to information delivery that consumes only the level of attention it deserves - in most cases it should be possible to provide information while you are performing other tasks.

This is achieved by taking a Personal Relevancy value and using it to make a presentation decision.

Is it background noise? Perhaps it should only form part of a passive backdrop (i.e. scrolling across a news ticker). Perhaps it is informative and deserves display on the bottom right of the screen so you can glance at it if you’re free. Perhaps it is time-sensitive and should follow your mouse around for a few seconds to make sure it catches your eye. Perhaps it is mission critical and requires your full attention and interaction.

In this way, Measured Disruption is about intelligently differentiating incoming information and presenting it in a full spectrum of attention appropriate formats.

This type of Attention Allocation is at the heart of the Touchstone Attention Management Platform. Its input adapters collect information from various sources (that could include explicit subscriptions, personal recommendations and activity based recommendations). Its Personal Relevancy Engine determines the level of importance of each item, and it invokes the right output adapter(s) for presenting the information to the user. Presentation styles vary based on the importance of the information.