Not long ago I wrote about Engagd, the attention service where you can view content from any site or feed, based on your lifestream feed, clickstream feed and/or any URL that represents your likes, preferences, etc…
This also has a sister service called particls, the gist here, is this RSS alert system filters/ranks content (from your subscriptions according) to your past reading behaviour, this also determines the delivery method for this content (there’s all sorts of manual override).
I have posted about attention in general, just like R/WW latest attention post on the attention economy, but it’s time to get to actual experimentation…this is a long time coming…my first post on attention was 21/11/2005.
Types of Attention
APML is a container that stores your attention, when you interact with a site or with a sites feed, there is the potential for your attention to be captured and output as an APML file, it can also be output as an RSS feed.
Another way to capture attention is very easy; what you blog about, what you bookmark…if you go to the trouble of publishing something and bookmarking, it most probably means you like it, unless you dedicate your time to talking about stuff you don’t like, in that case you have another issue, ie. paying attention to stuff you don’t like…you may stop and think to dedicate your time to stuff you do like. Looking at your past attention says something about you as a person.
NOTE: Actually that’s interesting, someone may pay attention to particular polititcians they don’t like, and publish stuff as a way to inform the public, this is noble…so, in a way their attention is focused on what they don’t like, but it is their attention just the same.
I guess it’s more about the results that come from your attention.
Anyway, here are some types…
1. RSS Reader - top ranked feeds based on your reading habits (output as APML)
* this could be output by an SLE feed (keep updated to a list of the current Top 10 feeds by a user)
2. RSS Reader - new top ranked items based on your reading habits (output as RSS)
* this is new ranked items your RSS Reader thinks you will like based on your past reading habits and actions (demonstrated by clicking items to read, saving, time spent, rating, etc…)
An important part of this is ratings, as you are explicitly telling the system if you like an item or not, so even though you gave it attention, this indicates the quality of the attention.
Just as important is that it records what you don’t like, what you rate negative, or what you choose not to click (read) or save, etc…
* to be clear this feed does not include past data from your actions, it’s just newly ranked items based on your past behaviour. This means that you could grab the RSS feed of a particls user and see the world as they do. RSS Readers like Rojo have a general RSS feed, meaning you can subscribe to a river of news of another person’s whole RSS Reader. What makes Particls different, is that this river of news is ranked content that has been personalised to that users reading habits. With Rojo you can see all posts that are user has in their RSS Reader, whereas in Particls you are seeing these posts in a personalised order, so you are even moreso getting an insight into another users world.
3. Lifestream - new items based on your actions, such as publishing a blog post, bookmarking an item, etc…this is based on your blog feed, bookmark feed, etc… (output as RSS or APML)
* point 2 is the system giving you stuff in a ranked order it thinks you will like to; read, blog about, bookmark, etc…whereas point 3 is stuff you have already actioned…so they are kind of similar, only point 3 is more defined or complete (as you have judged it).
4. Clickstream - new items based on sites you have surfed (output as RSS or APML)
* surfing to a site doesn’t mean you like it, but if it can record implicit attention, like, the amount of time and number of times you spend at a site, then this is more holistic…rating sites as you surf is an explicit and accurate way to record attention.
particls
Import some feeds, or generate an inhouse keyword search feed, then filter/rank items by manual settings, such as keywords, and/or let items appear according to your past reading behaviour (attention).
Obviously the more you use this, the more it gets to know you, and the more it can rank stuff to what you only want to see.
So this is newly ranked items, filtered against how you have used this service in the past…particls will generate an APML file for you.
This does not contain any feed items, it only contains a list of your interests, and it may also contain a list of feeds you subscribe to and how much you are interested in them but never actual feed data.
Perhaps this is similar to Google Reader Trends.
Particls will also generate a highest ranked items RSS feed which you could input into Engagd Profiler as one of the feeds that make up your APML file.
So what would be the difference between these 2 APML files?
NOTE: these APML are both mentioned in points 1 and 2 respectively in the Types of Attention section above.
The first one has some interest keywords and highest ranked feeds according to your reading trends.
The second one has highest ranked new items (this is a more granular level, as we are talking about actual personalised items, and not just default all items from a feed, like the first one)
1st APML - Ranked feeds
If I browsed and uploaded the first APML file into Engagd Profiler, then in Engagd Item.Rank whack in the feeds I read, this would be weird, as the feeds I read are already in my particls APML file, I would just be comparing the same feeds to themselves.
Instead if I whacked in a whole different set of feeds into Engagd Item.Rank, then it would be displaying content from these other feeds based on the feeds in my APML file.
I guess this is like saying, I like the content from these 10 feeds (from my Engagd Profiler APML file, which came from my particls APML), give me ranked posts from these 10 other feeds (Engagd Item.Rank) that are similar to typical content from the 10 feeds in my APML.
As mentioned above Google Reader Trends could generate a similar APML, we could then use it as an attention file to compare against a new set of feeds or even a library of data like Amazon.
2nd APML - Ranked items
Now if we go to the second APML file (based on the generated highest ranked new items RSS feed from particls)…if I whacked in 10 new feeds into Engagd Item.Rank, this would return posts matched against actual items I like from subscriptions, this is more granular.
NOTE: Item.Rank does not return results, it makes various filtered feed versions of the feeds you entered, you can subscribe to these filtered feeds, which in inturn will show new items (results).
This is like saying, I will probably like the ranking of these new posts (contained in my APML file - based on my RSS reading habits), give me ranked posts from these 10 other feeds that are similar to the posts in my APML.
Another scenario is using this APML file against the current feeds in your RSS Reader, with this first APML example we said this may be a bit weird, as that APML only contains feeds (not items), and what’s the use of matching the same feeds against themselves.
But with this second APML (which contains items) it can be useful to measure against the feeds we already read…what we are saying is, I like the ranking of the new items that I read from these 10 feeds, from now on when I read these same 10 feeds, rank/filter the content according to what I have liked in the past.
If you use a personalised RSS Reader, then you need not worry about doing this, as it already does it for you continuously, but another person could subscribe to these filtered feeds based on your world…kind of like selling or sharing your RSS reading attention for a set of feeds (your attention is based on these same feeds).
Just the same, you could also share your RSS reading attention for a set of feeds that are different to the set of feeds your attention is based on.
As mentioned earlier, Item.Rank does not return results, it makes filtered versions of a feed you submit…in this example it means re-subscribing to the feeds in your RSS Reader with the filtered version you have made. These filtered versions are based on your attention, so the gist is you will see all posts, but ranked.
These filtered feeds are always synching to your APML file, so if your interests change, your filtered feeds will tune to this and start delivering content according to your new interests.
eg. If you subscribe to 10 web 2.0 feeds, and you stop clicking on posts about “wiki”, but start clicking and saving posts about “sms” moreso than you did before, well then your filtered version of these 10 feeds will receive this information and start delivering “sms” type posts to the top of your pile, and “wiki” type posts lower in the ranks.
NOTE: Instead of seeing posts ranked, you could choose to completely filter out stuff, but using your attention file in this way in this scenario wouldn’t be wise, as you would seldom see “sms” posts, as there would be no opportunity in seeing different stuff that would result in a change of interests.
This second APML file (new top ranked items) is kind of similar to your lifestream feed, only sort of…your lifestream feed is made up of items you publish and bookmark and even surf (clickstream feed), whereas the second APML file is also made up of items, but only high ranked new items in your Particls system, these are not articles you have actioned, eg. blog posted, bookmarked.
At the moment Particls has the new items ranked feed, but it will soon have feeds for actioned content, ie. feeds for rated items, flagged items, marked as read items (all part of its Pebbles feature)…but if you have a lifestream feed, then you already would be flagging stuff eg. bookmarking in del.icio.us, Google Reader Shared Items, etc…
NOTE: particls allows you to bookmark items into social bookmarks like del.icio.us, etc…
This feature is built into some personalised RSS Readers like Attensa, and Feeds2.0. It records your past reading behaviour (what you click, flag, save, rate, etc..), and then begins to show you content from your feed subscriptions in a different order, an order that should be pleasing to you, as it’s based on your past preferences.
If these types of RSS Readers output this clicked based attention information as a feed we could convert it to an APML file, better still it could just be output as an APML file.
It would be good if we could own this attention as we could use this file against other content outside of your RSS Reader.
eg. plug it into a library of data like Amazon or use it as a filter on a batch of feeds that someone gives you in an OPML file, instantly you are reading contents from these feeds according to your attention file (based on your RSS reading habits)…choosing the option to see all posts ranked, or to filter out posts you don’t like so you don’t have to see them.
So this is different than an RSS Reader having a new ranked items feed, this is moreso a file on your history of clicks, saves, etc…so it’s kind of like a clickstream, and lifestream feed all within your RSS Reader environment.
NOTE: to be clear, a new ranked items feed is new ordered items based on past clicks, whereas a click history feed/file contains these actual past items you clicked on.
engagd
So how do you use APML?
What I recommend is get yourself a lifestream feed (stuff you post, bookmark, etc…), this way your APML file will have granular items…you must like these items as you blogged about them and bookmarked them.
NOTE: I guess some sort of word/link frequency would determine how much you like some items over others…not sure.
I mean how can it determine the aboutness of which blog post you liked writing about over other blog posts…maybe it could look at your categories/tags in your blog and bookmarks and decide that since this tag has more posts/bookmarks you must like it more.
But then again you may have a tag that contains only one bookmark, and this may be your most liked bookmark…I guess this is where manual rating is handy.
Anyway, that’s just more on the depth of your attention juice, I’ll leave that up to the techies.
OK, so you input your lifestream feed into Engagd Profiler, if you don’t have a lifestream feed I guess you can kind of make a backend one at Profiler, as it allows you to enter multiple feeds, just enter one for your blog, bookmarks, etc…
Then it will generate you an APML URL…I haven’t shopped this APML around yet, ie. plug it into a website or library of data, in order to have an immediate personalised experience, this is what we hope to do soon as more websites APMLify.
But for the moment what I can do is goto Engagd Item.Rank and enter a feed/s from a website/s I want to check out, this will in turn generate filtered versions of this feed/s, which I can then subscribe to in my RSS Reader.
Example of various filtered feeds generated from a feed filtered against an APML
eg. SUBSCRIBE: Full Ranked Feed | Most Items | Important Only | Very Important Only
What this means is I have made filtered versions of these feed/s based on my APML file (which is based on my lifestream feed, and any other feeds).
The attention experience
As mentioned earlier if I could visit Amazon directly and plug in my APML, it will list me stuff from its catalogue that it gathers I will like, even if I have never visited the website before…this is the ultimate personalisation scenario.
Now what about if I want to continue doing this for new items added to Amazon (if Amazon had a latest items RSS feed, it does for user tags), I could filter this feed based on my APML file, this way I don’t have to visit Amazon everyday, see the example below.
NOTE: In the examples below I use Amazon as an example, another obvious choice is to use del.icio.us latest items RSS feed.
NOTE: I have not tested these examples below, but anyone can start testing now, we have the tools. I guess my next post on this topic would be some test cases and results to see if the practical meets the theory.
eg. 1
Enter your lifestream feed into Engagd Profiler
Enter Amazon latest items feed into Engagd Item.Rank
Now you will have a choice of new filtered versions of the Amazon feed to subscribe to in your RSS Reader
SUBSCRIBE: Full Ranked Feed | Most Items | Important Only | Very Important Only
Subscribe to the “Most Items” version of the Amazon feed into your RSS Reader.
How does this differ than subscribing to the regular Amazon feed?
It will show you only those new items (Most Items) from Amazon based on your lifestream (stuff you blog about, save….)
How about that, a personalised experience without ever having used Amazon before…imagine doing this against the many library catalogues and journal databases, etc…
NOTE: this is going further than ranking items, you are actually filtering items out
eg. 2
Enter your 10 favourite blog feeds into Engagd Profiler
Enter Amazon latest items feed into Engagd Item.Rank
It will churn out a choice of 4 filtered feeds for each feed you entered, if you choose the “Full Ranked Feed”…
it will only show you all new items (Full Ranked Feeds) from Amazon, but ranked according to the content from your 10 favourite blog feeds.
NOTE: Full Ranked Feeds is the full feed with rank tags for each item
eg. 3
Enter your clickstream feed into Engagd Profiler
Enter Amazon latest items feed into Engagd Item.Rank
It will churn out a choice of 4 filtered feeds, if you choose the “Important Only”…
it will only show you those new items (Important only) from Amazon based on the content from your clickstream feed (sites you surf)
eg.4
Enter your RSS Reader most ranked new items feed into Engagd Profiler
It will churn out a choice of 4 filtered feeds, if you choose the “Full Ranked Feed”…
it will show you those all items (Full Ranked Feeds) from Amazon, but ranked according to your RSS reading habits, ie. typified by items you usually click, read, save, etc..
NOTE: particls is the only RSS system I know of that delivers a most ranked items feed, other personalised RSS Readers choose not to output this data.
eg. 5
Enter your lifestream, clickstream, and most ranked items from your RSS Reader as your APML file
….you know the rest.
APML and OPML
What if I upload my lifestream feed as my Profiler APML file.
Then I enter my RSS Reader OPML into Item.Rank (something you can’t do yet, so you have to enter each feed manually).
Now I will have a choice of filtered versions of each feed from my OPML.
If my OPML has 50 feeds, I could now select which filter I want on each eg. feed 1 Most Items, feed 2 Very Important Only, feed 3, Full Ranked feed, feed 4 Most Items, etc…
When I’m done I hit generate and it would make me a new OPML URL (hopefully we will see this feature soon)
Then I go to my RSS Reader and delete all my feeds, and enter my new OPML.
Isn’t this nice, OPML and APML being used together 
NOTE: actually just enter your new OPML into a new RSS Reader to see if you like it first.
Now when I read my RSS Reader, I only see content I want to see with some feeds, and other feeds I may have chosen to see all content ranked (all this is based on my lifestream)…this is the concept.
I suppose this idea is similar to what Attensa would automatically do for you.
As mentioned before Attensa will track my reading behaviour, and rank items in my stream according to my attention, it may also choose to not show me certain stuff at all (as I never click on these type of posts)
eg. If I never click on “Second life posts”, but I always click on “Google” posts, my RSS Reader will be tuned to always rank posts about “Google” on the top of my pile as it know I like them…and it will rank posts about “Second life” on the bottom of my file as it knows I don’t pay attention to these posts, it could even take them off my radar all together.
All these actions I’m performing in a personalised RSS Reader like click, flag, save, rate is demonstrating my attention within this system, similar to how my lifestream/clickstream show my attention.
The main difference is my lifestream is mine, and not restricted to one service, I can take it away and apply it to another set of feeds or pool of data.
RECAP
What makes up attention?
- Past RSS reading habits feed or file…contains actual items you have clicked (a combination of a clickstream and lifestream limited to the RSS Reader arena)
- A feed for future RSS Reader items ranked based on past reading habits
- A list of ranked feeds (not items), based on your past reading habits
- Lifestream feed
- Clickstream feed
How can I convert these attention captured feeds in a usable format?
- Create an APML file at Engagd Profiler
Then what?
1. Plug your APML file into a library of data like Amazon or a journal database (this is what we hope to do soon…ie, visit a website and get a personalised version of that site)
2. What we can do for now is if this website/s has a feed, we can then make a filtered version of this feed to see content ranked to our preferences, or even filter out content all together.
This option 2 saves you visiting the website all the time, like in option 1.
3. If your RSS Reader doesn’t have a personalisation feature, you can create filtered versions of your feed subscriptions against your lifestream, and then subscribe to these feeds as a replacement.
![email.png](http://services.nexodyne.com/email/icon/GxbiQvgbwBgCnw%3D%3D/oIp%2BJMQ%3D/R01haWw%3D/0/image.png"){kind=icon}
