“If you want to inform yourself of the basic principles of attention profiling or need to explain the concept to others then please read on. Feel free to add your clarifications, your
conclusions and your constructive criticism to this deliberately non-geek conversation.”
In recent months quite a few bloggers covered the growing adoption of APML, a proposed standard for attention profiling. Those about to give up reading here already, please don’t. I personally found most of these posts delving in rather deep. If you want to inform yourself of the basic principles of attention profiling or need to explain the concept to others then please read on. With today’s post I’d like to make an attempt at writing a layman’s article answering exactly these three questions:
- What is attention profiling and what are the benefits?
- What tools and services already support or endorse attention profiling?
- Where could you go next?
As usual, this post concludes with a news radar.
I encourage you to participate in this deliberately non-geek conversation about
attention profiling, either by posting a comment or by writing a blog
post of your own. Feel free to add your clarifications, your
conclusions and your concerns.
I like introducing attention profiles
as consolidated, structured descriptions of people’s interests and dislikes.
The information about your interests and how much each means to you
(ranking) is stored in a way so that computers and web-based services
can easily read it, interpret it, process it and pass it on should you
request and permit them to do so.
In today’s post I confine myself to describe services that are capable of handling attention profiles based on the proposed APML standard. To
make it easier for humans to recognize that files containing an
attention profile indeed are attention profiles, we label them with the
file extension “.apml”.
Technically attention profiles can be stored, accessed and updated from multiple devices and multiple operating systems.
In February 2007 Web 2.0 industry analyst Emily Chang sparked an avalanche of comments with her blog post My Data Stream. Looking back, Emily’s post could be considered the tipping point that led to many online conversations about attention, spawning new technology start-ups promising to solve the problem Emily described. The first few sentences from her post:
“As the calendar rolled to 2007, I kept wishing I could look at all my
social activity from 2006 in context: time, date, type of activity,
location, memory, information interest, and so on. What was I
bookmarking, blogging about, listening to, going to, and thinking
about? I still had the urge to have an information and online activity
mash-up that would allow me to discover my own patterns and to share my
activity across the web in one chronological stream of data (to start
I strongly suggest you read Emily’s post in full. The nice thing was that Emily kept updating her post with links to related items as they appeared. Don’t skip the ensuing conversation in the comments, as it is still alive today.
Emily recently chose Engagd to store her attention profile. Click on the image to open the full profile:
Attention profile sources
Though not an exhaustive list, your attention profile could be based on:
- pages you bookmark and tags you assign
- your favorite videos, music and TV shows
- hyperlinks you follow and share with your friends
- things you write about and topics you keep track of
- items you click on in your feed reader
- things you buy from a web store
- places you visit and events you attend
Purpose of attention profiling
Attention profiling has multiple aspects. To me the most important
aspect is that it allows me to reduce my information fatigue
considerably. If I have enough time on my hands I’m still free to
wander off on the web and jump from one page to the next. However, in
the usual circumstance that my time is limited and my attention is
scattered between tasks competing to top my to-do list, my attention
profile helps me to focus on my core interests. Prerequisite of course
is that the services and tools I spend most of my time with are aware
of the existence of my attention profile.
The privacy aspect of attention profiling is brought up quite
often when I talk to people. They consider their APML as their private
property and are usually afraid their browsing behavior will be exposed
to prying eyes. I look at the privacy aspect of attention profiling
from a different angle: right now sites like Facebook and Google
collect usage data from and about me. They know about my interests,
they know what sites I open and they know who my friends are. At the
moment all this is a one-way operation: they collect the data that I
give to them and I get no insight as to how they filter the content they or their advertisers offer to me. I prefer to have that information distilled into an
attention profile so that I can at least have control over whom I share
this information with.
Of course there’s a commercial aspect to attention profiling as well: most people know and appreciate Amazon’s recommendation engine. Actually this mechanism is Amazon’s proprietary attention profiling system based on the product pages you open and the items you put in your shopping cart. Imagine being able to take this profile with you onto other web sites that are
capable of producing dynamic content based on your interests. So it’s not just
about niche targeting of advertisements, but also, for example, about removing
sports pages from the home page of my favorite news headlines site once it
becomes clear I’m not interested in soccer and baseball.
Building your own attention profile
The smartest way to
get started having your own attention profile is to sign up with
attention-profiling service Engagd.
Engagd is a site where you can have your attention profile generated, updated
and dynamically hosted.
The direct URL to my profile is https://apml.engagd.com/apml/cleverclogs.is.engagd.com
After signing up with Engagd, you’ll be taken to the Engagd Profiler, a page where
you can specify from what kind of attention data Engagd should generate your
attention profile. I’ll be explaining each of these straight away. The Engagd
Profiler can process these types of files:
- individual web pages
- RSS feeds and OPML files
- Life stream feeds (technically RSS feeds)
- APML files
A single web page that just happens to reflect your
interests fairly well suffices to build an APML file. Just provide the URL of
that page and Engagd will build an attention profile from it.
RSS feeds and OPML files
The Engagd Profiler can process
individual RSS feeds. Just paste the URL of your RSS feed into the input box. I
found out that the Engagd Profiler works even better if you provide it with the
URL of a list of RSS feeds (OPML).
Life streams are a slightly more advanced, but also a
considerably more accurate way to start your own attention profile. In most
definitions of life streams they refer to a continually refreshed stream of the
bits and pieces that constitute your personal interests. Life streams are
usually generated by combining updates from multiple RSS feeds. Some people’s
life streams are simple, containing just the items they select in their feed
reader and the bookmarks they create. Other people’s life streams can turn out
to be complex and voluminous, depending on their productivity. Mine is somewhere
in between: you can see it in action in the black widget above the fold here on
CleverClogs. I named my life stream CleverJots and its contents
fluctuate while I sign up to try out new services.
Here’s how I added my life stream to the Engagd Profiler. The URL that you
see in the input box is generated by the service that I used to create my own
life stream: FeedBite. FeedBite lets you
combine multiple feeds into one. It also provides you with an OPML file that
they host for you. It’s the URL for this OPML file that I pasted into the input
The reason I submitted my OPML file and not the single CleverJots feed is
because the Engagd Profiler turns out to be better capable of distilling my
interests from the individual feeds constituting an OPML file (categories and
tags still being intact), than from the combined feed.
Here are the feeds that the Engagd Profiler discovered in my OPML file:
As you can see from this screenshot, all of the feeds included in my profile
are releated to my activities on the web: content I create, comments I write,
links that I share. The more personal the feeds you offer to the Engagd
Profiler, and of course the more items those feeds contain, the better it can
distill your interests.
There are quite a few services that let you create a life stream. From rather
simple, wizard-led websites to full-fledged newsmastering services, most of them
create a custom feed or an OPML file for you that you can use as the basis of
your attention profile.
To learn more about life streams and how to create them, check out John
Tropea’s in-depth coverage of all services in this market in his post
about life streaming services over on his commendable blog Library Clips. What I like about
John’s reporting style is that he puts each service within its context: what
other services are there, how could you benefit from each of them and what
potential does each of these have.
I’d also like to point to the Google Blog Search feeds that I included in
this post’s news radar. You’ll find the radar at the bottom of this post.
APML files – it makes sense that a service that stores your
APML file for you is also capable of importing one, so that’s not too difficult
to understand that the Engagd Profiler can import those directly, I presume.
So, where would you go to obtain or generate an APML file?
Well, the number of tools and web services that actively support APML is growing
steadily. APML awareness means that developers add functionality to their
product so that it can import an APML, enrich it with personalized attention
data generated by their own product and pass that information on to other
products aware of APML.
Among the web services that have jumped on board the APML train are Dandelife and Cluztr. From recent insinuations on the
Bloglines blog, it is expected that Bloglines will soon join the ranks too. If
you’re a Windows user, then you may want to check out Particls, a desktop program that turns
keywords and feeds into a continuous display of relevant news items.
A growing collection of questions and
answers about attention profiling in general and the Engagd service in
particular can be found on the Engagd FAQ wiki
If you would like to know more about the APML specifications and track their
development, I suggest you start out by signing up for the fast-growning public APML discussion group on Google
Groups. Just click on the link and choose the Join option. You can actively
participate or just lurk for a bit and see what kind of topics other members
come up with. A thread I find particularly helpful is the “introduction” topic,
where members are encouraged to introduce themselves and explain from what angle
they have joined the group.
APML on facebook
Another option is to sign up for the APML group on
Facebook. As of this post’s most recent update (2007-10-28) it has 268
An open standard, APML is the subject of lively debate among the members of
the APML Workgroup (click to see
a current member list). Starting point are the APML specs. If you believe that you qualify
for Workgroup membership (vendors-with-plans,
XML-specialists-with-constructive-contributions), then write to Chris Saad
directly or start out with the public forums on Google Groups and facebook
mentioned in the previous paragraphs.
Attention Profiling – Roadmap for the nearest future
quite safe to assume that Web 2.0 service vendors are closely watching this
space. Some are waiting for the big players to adopt the technology, others
prefer to lead the way themselves. I personally would not be surprised at all if
notable players such as del.icio.us, Clipmarks, StumbleUpon, Digg and web page annotation services such as Diigo and Fleck would become APML aware.
Then of course there’s the group of RSS tool vendors and feed aggregators. I
have quite high expectations from vendors of top-notch products such as FeedDemon, NewsGator Online and BlogBridge. The Google Reader developers may find
that implementing support for APML fits their strategy of streaming additional
personalized content and advertisements to their enormous user base.
Further reading (updated 2007-12-30)
Quite a few solid
posts have been written since I wrote mine. The announcement by NewsGator
Technologies, Inc that they’ll implement APML throughout their product range
(and likely their API) is definitely a highlight. Here’s a selection of the ones
you might be interested to read:
APML: what it is & why you want it (by Elias Bizannes, 2007-10-10)
NetNewsWire and NewsGator Inbox to Support APML (by Nick Bradbury,
NewsGator and Bloglines Join APML Workgroup (Digg
buster by Marshall Kirkpatrick, 2007-10-15)
support for APML this time from NewsGator one step closer to the Enterprise
(by Daniela Barbosa, 2007-10-15)
with APML support (announcement by Adam Kalsey, developer of Feed Crier and
implements APML: the value of standards in an open world (by Ross Dawson,
Considerations on Google Reader and APML: How
Google Reader can finally start making money (by Elias Bizannes;
have your attention? (by Brad Feld, 2007-10-16)
APML and MiniBar (busy week) (by Nick Halstead, 2007-10-21)
to APML (by Emily Chang, 2007-10-21)
APML: The Next Big Thing or the
Next FOAF? (by Mark “rizzn” Hopkins, 2007-10-22)
is lifestreaming and attention – part one (by Piers Jones, 2007-10-24)
Attention Profiling & APML
(by Elliot Turner on Aqua Regia, 2007-11-02)
Profiling: APML Beginner’s Guide (by Michael Pick on Master New Media,
goes into private beta (by Aaron Tan on Confessions of a Technophile,
Should BlogBridge support APML ? (by Pito Salas, CEO of BlogBridge, 2007-11-15)
Explaining APML to Suda (by Ian Forrester on Cubicgarden.com, 2007-11-18)
An APML Experiment (by Paul Lamere, Sun Labs researcher, 2007-11-21)
APML Support for WordPress (by Matthias Pfefferle, 2007-11-28)
APML for Del.icio.us (by Paul Lamere, Sun Labs researcher, 2007-11-30)
User Centric design and identity with Beacon (by Uno de Waal – South-African blogger – 2007-11-30)
Coverage in other languages
In Dutch: Attention
profiling: ja graag! (by Tibor Paulsch, 2007-10-12)
In Italian: Attention
Economy: Bloglines e NewsGator adottano l’APML (by Enrico Bertini,
In Portuguese: APML:
uma forma de dizer ao mundo o que merece sua atenção (by Bruno Torres,
In Spanish: APML,
crea y controla tu propio perfil de atención (by Alvaro Castaño,
In French: Qu’est ce que
l’APML: Attention Profiling Markup language (by Julien Grière,
In Japanese: ?????????????????
APML ?? (2007-10-22 on Social Networking.jp)
In Greek: APML
- Attention Profiling Mark-up Language (by Stelabouras on Wigger.gr,
In French: APML
- Attention Profiling Markup Language (by Mathieu Favez, 2007-10-27)
In German: APML
- Sag mir wer du bist und ich sag dir was dich interssiert (by Cornelius
In Italian: Le
Metriche attenzionali e l’APML: uno strumento per misurare l’autorevolezza sul
web? (by Sacha Monotti on MediaMeter, 2007-10-30)
In Swedish: Är
framtidens sociala sajter asfalterade med OpenSocial och APML? (by Fredrik
Wass on bisonblog, 2007-11-07)
In Lithuanian: APML – prekyba
interneto sielomis (by vienastoks on nežinau.lt, 2007-11-13)
In Russian: APML – ???? ???????? ??????? ???????? (translation of Michael Pick’s post by Daniil Bratchenko, 2007-11-21)
In Spanish: APML y el perfil de interés del usuario(by Antonio Cordiz, 2007-11-21
In German: APML – Attention Profiling Mark-up Language (by Matthias Pfefferle, 2007-11-23)
In Russian: APML – ????? ?? ???????? ???????? ? ??????? (by Svetlana Gladkova, 2007-11-30)
In Swedish: Låt APML bli en genväg till användarna (by Björn Fant, 2007-12-28) – This article was also published on the online edition of the Swedish MacWorld (also by IDG).
In French: APML, un profil dont vous êtes le héros (by David Larlet, 2007-12-30)
Attention Profiling Grazr
I’ve put together a quick
reading list that you can copy if you like. The top feed aggregates items from
the other feeds with the most recent item appearing at the top of the
And by request: