Eleet.fi suggests links based on what you’ve shared on Facebook

(Cross-posted from the Eleet.fi blog)

Hello and welcome to Eleet!

Eleet suggests links to you that you might be interested in, based on links that you and other people have shared on Facebook.

Please, have a look. The more active you are at sharing links on Facebook, the more interesting links we will be able to suggest to you.

‘Eleet’ is Finnish for ‘gestures’. http://eleet.fi is the first public link recommendation service built on a recommendation engine that is being developed through a project called ‘Project Gestures’.

Facebook is the first social web service from which we source shared links – or ‘gestures’ in our jargon. We plan to roll out similar integration with other major services on the Social Web, next up Twitter. Also, we want to enable you to submit any blog or RSS feed as a source of your on-line gestures. Just watch us 🙂

We will publish the source code of the engine’s core technology during September 2011.

During the spring of 2011, our project received recognition as it was selected to the finals of the first-ever ‘Uutisraivaaja’ (‘Newsplorer’) innovation contest sponsored by the Helsingin Sanomat Foundation.

The Foundation is linked to Helsingin Sanomat, one of Finland’s most influential daily news papers. Inspired by the Knight News Challenge in the United States, the contest “(…) seeks ideas for improving and renewing the distribution of information. (…)”

Our project was in part inspired by a tweet from New York University Journalism Professor Jay Rosen, who wrote while linking to one of our earliest blog post on the topic:

“Imagine a personal recommendation system for news based not on consumption habits but on your gestures: authoring and sharing.”

So, the fundamental underlying principle of our engine is that we don’t analyze what you click and read on-line, but what you write and share. When you share a link on Facebook, Twitter or on your blog, that is a very strong indication, or ‘gesture’, that the topic behind the link is relevant to you.

Well, ‘nuff said. Please, take our engine for a spin a let us know what you think!

Hello and welcome to Eleet!

Eleet suggests links to you that you might be interested in, based on links that you and other people have shared on Facebook.

Please, have a look. The more active you are at sharing links on Facebook, the more interesting links we will be able to suggest to you.

Background

‘Eleet’ is Finnish for ‘gestures’. http://eleet.fi is the first public link recommendation service built on a recommendation engine that is being developed through a project called ‘Project Gestures’.

Facebook is the first social web service from which we source shared links – or ‘gestures’ in our jargon. We plan to roll out similar integration with other major services on the Social Web, next up Twitter. Also, we want to enable you to submit any blog or RSS feed as a source of your on-line gestures. Just watch us 🙂

We will publish the source code of the engine’s core technology during September 2011.

During the spring of 2011, our project received recognition as it was selected to the finals of the first-ever ‘Uutisraivaaja’ (‘Newsplorer’) innovation contest sponsored by the Helsingin Sanomat Foundation.

The Foundation is linked to Helsingin Sanomat, one of Finland’s most influential daily news papers. Inspired by the Knight News Challenge in the United States, the contest “(…) seeks ideas for improving and renewing the distribution of information. (…)”

Our project was in part inspired by a tweet from New York University Journalism Professor Jay Rosen, who wrote while linking to one of our earliest blog post on the topic:

“Imagine a personal recommendation system for news based not on consumption habits but on your gestures: authoring and sharing.”

So, the fundamental underlying principle of our engine is that we don’t analyze what you click and read on-line, but what you write and share. When you share a link on Facebook, Twitter or on your blog, that is a very strong indication, or ‘gesture’, that the topic behind the link is relevant to you.

Well, ‘nuff said. Please, take our engine for a spin a let us know what you think!

Advertisements

A recommendation engine informed by gestures

[UPDATE, June 13, 2011: We’ve activated the eleet.fi domain and blog.]

(Cross-posted on the Uutisraivaaja blog)

We consider the above tweet by New York University Journalism Professor Jay Rosen to be the most flattering endorsement to date for the idea of ‘Gestures’, or ‘Eleet’ in Finnish.

Our vision is to create a personal recommendation engine informed by on-line social gestures.

Wake up your favorite glowing rectangle and your Internet browser, RSS reader, Facebook or Twitter client, or special-purpose mobile application will show you hyper-relevant Internet destinations.

The way in which we interact with on-line information is informative of our news preferences.

Find me stuff I’m interested in.”
Dave Winer

Subscribing to an RSS feed; reading, storing, sharing, tagging, rating or sending an article; writing a blog post, commenting on or linking to one: all these actions can be seen as ‘gestures’, indicating that the user attributes some degree of relevance to the content at hand. Continue reading

Elevator pitch for #Uutisraivaaja ‘Gestures’

Just to capture this for future improvement: I was asked by Tanja Aitamurto, coordinator of the Helsingin Sanomat Foundation‘s Uutisraivaaja contest, to send in an elevator pitch for my entry, ‘Gestures’.

Gestures is one of the ten ideas selected to the Finals of the contest (out of 257), to be held in September.

I didn’t have much time to write the pitch, so here it comes, quick-and-dirty.

Borrowing from my blog post at http://josschuurmans.com/2011/03/21/picking-your-brain-your-skill-your-network-and-your-money/ and from my Uutisraivaaja application, the pitch goes something like this:

My innovation idea for the Uutisraivaaja contest is called ‘Project Gestures’.

News organizations need a better way to tailor their news and information to individual readers/users. A sophisticated way of building a personalized news offering is through the analysis of on-line social news gestures.

Gestures are ways in which web users interact with information that they encounter, thereby indicating relevance. Gestures may include: subscribe, read, store, share, tag, rate, copy-share, send, comment, blog, micro-blog, pipe-through, link, approve/reject.

Every gesture contributes to a collective human news filter. The aggregate data can be used to inform a personal news offering to individual users.

For every pair of users in the system, the engine will calculate the proximity of their past gestures.

If user A and user B have a high proximity of past gestures, the next gesture user A will express will be highly relevant to the engine’s recommendation of news and information to user B; and vice versa.

The personalized news headlines can be offered as a stand-alone service, or can be integrated in existing on-line news outlets.

Borrowing your brain, your skill, your network

What follows is a draft email message to people in my network. If we know each other, move me to your spam whitelist and expect to receive one of these 🙂

[STARTS]

Subject: Borrowing your brain, your skill, your network

Hi!

I’m on cloud nine at the moment because of the recognition I received from the Helsingin Sanomat Foundation’s ‘Uutisraivaaja‘ innovation competition for my idea of a personal news recommendation engine based on on-line social gestures.

257 entries where submitted to the competition. 10 ideas have been selected to the Finals. I will be given 10,000 euros to prepare a presentation of my innovation at the Finals on September 15.

In a nutshell (from my application):

Gestures are ways in which web users respond to the information which they encounter, thereby indicating relevance. Gestures may include: subscribe, read, store, share, tag, rate, copy-share, send, comment, blog, micro-blog, pipe-through, link, approve/reject.

Every gesture contributes to a collective human news filter. The aggregate data can be used to inform a personal news offering to individual users.

For every pair of users in the system, the engine will calculate the proximity of their past gestures.

If user A and user B have a high proximity of past gestures, the next gesture user A will express will be highly relevant to the engine’s recommendation of news and information to user B; and vice versa.

While I’m not yet entirely sure how to go about it, here are some initial thoughts:

  • Create an open source software project which may outlive the duration of the competition;
  • Establish a foundation to manage and support the project and safeguard its open source nature;
  • Engage a Chief Engineer;
  • Engage a Chief Designer;
  • Engage a Chief Evangelist;
  • Engage logistical and other support people;
  • Engage (international) news organizations to sponsor the project;
  • Engage non-commercial organizations in the realm of news and journalism;
  • Engage educational institutions;
  • Explore governmental, non-governmental and philantropical funding;
  • Explore venture capital funding;
  • Build the core technology;
  • Design a compelling web/mobile user interface;
  • Design a convincing demonstration of a common use case scenario;
  • Build a browser plug-in;
  • Launch a web service featuring personal news recommendations;
  • Initiate pilot projects with one or more news organizations;
  • Nurture users and developers;
  • Give sponsors, developers, and anyone else who adds value to the project due credit on the website and elsewhere.

Okay, so why am I writing this to you? Because I’m taking the liberty of borrowing your brain, your skill, your network and/or your money. Specifically:

  1. What is the first thing that comes to your mind when reading all this?
  2. What do you like/dislike about the idea?
  3. What is the best advice you can offer?
  4. How would you go about securing success in the Finals?
  5. How would you budget the 10,000 euros?
  6. (How) would you like to contribute to this project?
  7. To whom would you turn for sponsoring, partnering, funding?
  8. Whom else should we get involved?
  9. Any funding models / business models that I have overlooked?

Any response is highly appreciated, either in the comments or by email to <jos@josschuurmans.com>.

In case you feel like reading up:

‘Gestures’, my idea for a social news recommendation engine, got further in the Helsingin Sanomat Foundation’s competition!
http://josschuurmans.com/2011/03/18/gestures-my-idea-for-a-social-news-recommendation-engine-got-further-in-the-helsingin-sanomat-foundations-competition/

My submission to Uutisraivaaja #fortherecord
http://josschuurmans.com/2011/03/09/my-submission-to-uutisraivaaja-fortherecord/

A hierarchy of gestures for the Holy Grail
http://josschuurmans.com/2011/02/18/a-hierarchy-of-gestures-for-the-holy-grail/

Booting up a personal recommendation system for news
http://josschuurmans.com/2011/02/18/booting-up-a-personal-recommendation-system-for-news/

Transcript of 9 minutes ‘Rebooting the News’, episode 82
http://josschuurmans.com/2011/02/17/transcript-of-9-minutes-of-rebooting-the-news-82/

[ENDS]

‘Gestures’, my idea for a social news recommendation engine, got further in the Helsingin Sanomat Foundation’s competition!

How cool is this?

I received a telephone call from Ulla Koski this afternoon, informing me on behalf of the Helsingin Sanomat Foundation that my submission to the Uutisraivaaja competition was selected to the second round by a jury meeting earlier today.

In this second round of the competition, the Foundation will hand me 10,000 euros to further develop my idea, ‘Gestures’. The ten entrants to the second round are invited to Sanomatalo in Helsinki on Wednesday morning, March 23.

So I expect to have more details on Wednesday.

Via Uutisraivaaja-kilpailun jatkoon päässeet on valittu:

Uutisraivaaja-kilpailun jatkoon päässeet on valittu tänään 18.3.2011 tuomariston kokouksessa. Kilpailuun osallistui 257 hakemusta ja jatkoon pääsivät seuraavat 10 projektia:

  • Nicolas Kayser-Bril: Influence Networks
  • Jussi Pullinen, Lauri Eloranta, Aleksi Moisio, Petro Poutanen: Murut
  • Niko Lappalainen: Collapic
  • Johannes Koponen: Huome.net
  • Hanna Harilainen: Virtuaaliketju pellolta lautaselle+Jaana Kokkonen, Lilli Linkola: Uutiskoodi = yhdistetty yhteisprojektiksi
  • Annikka Mutanen, Susanna Niinivaara: Tutkiva uutispalvelu Huuhkaja
  • Torsti Schulz, Joona Lassila, Stefan Richter, Martin Richter: Naapurisopu 2020
  • Olli Sulopuisto: Oma radio
  • Jos Schuurmans: Gestures
  • Kimmo  Mäkilä: Faktat oikein

Voittajat julkistetaan Säätiöpäivänä 15.9.2011.

A hierarchy of gestures for the Holy Grail

After transcribing the pertinent 9-minute passage from RBTN 82 and offering some conceptual input to the idea of developing a personalized recommendation system for news, I kept thinking about the different kinds of gestures in the mix.

So it might be useful to establish a hierarchy of on-line gestures, which can serve as signals indicating endorsement or recommendation.

1. Subscribe. To a feed, a publication, a newsletter. Or even the repeated act of visiting a service or a web site. That’s a gesture saying: this is potentially interesting to me. Or, in some cases: I know that this is interesting to me.

2. Read. As Doc Searls might say, when I read something, it means that I let it inform me. I let it “author” me. I’m voluntarily exposing myself to its influence. I hope or expect to gain something from acquiring the knowledge or information encapsulated in the article or story.

3. Store (and tag) privately. Make it findable for myself. It builds an archive of things that I’ve read. I find it worth documenting that I read it. And I expect that sometime in the future I might find it worth retrieving it and re-reading it or using it some way or another.

4. Share. For example on Google Reader, as I tend to do with news. In addition to making it findable to me, sharing the feed publicly is also something of an endorsement or recommendation. Or at least, it communicates to anyone interested that I have read this and found it worth putting that fact on the record. Twitter, Facebook, Dig, Reddit, StumbleUpon…

5. Tag publicly. Contribute to the public goods of findability and folksonomy. (Same services as above)

6. Rate. Personally I don’t rate content much. What’s the point? What’s the benchmark? Except Facebook “likes”, which kinda combines rating, tagging and sharing.

7. Copy-share. Arguably it takes a slightly bigger effort to copy content into a draft blog post, although “Press This” makes it almost as easy as any other browser bookmark.

All public gestures of sharing and tagging are instances of amplification. And as we know, amplification is the new circulation. When sharing, two things happen. One: I help this piece of content which I find interesting, to find more readers, to get more exposure. Two: I endorse it, because I kind of associate my name with it when I tweet it or put it on Facebook or on my blog.

But I don’t necessarily interpret it. It can be: this is interesting, an endorsement, a recommendation for reading. Or the purpose may just be to say, I am reading this kind of stuff. My mind is now working with this kind of stuff. So, thinking about the edges of the social networks, if someone else reads the same, finds it interesting, then maybe it’s something worth talking about.

It’s a message from me, indicating that I’m open to conversation about this topic.

8. Send the article (link) to someone I know, for whom I think it may be highly relevant. The threshold for making this gesture is quite high. It’s a strong gesture, and it’s also a very personal gesture, not a public one.

9. Comment on a blog post or news article. Nowadays I don’t often do that, because I think that if it is worth commenting on, it’s worth keeping that comment on my own side, on my own blog, on my own “infrastructure (as I think Dave would agree).

10. Blog about the topic I read, because I have something to add: interpretation, commentary, fact, opinion, context. Or to take it in an entirely new direction. The point here is to create original content. It can be a blog post, a tweet, a status update or what have you. For our purposes, in order for this to be a gesture it’s important to link back to the original article/post/story.

11. Possibly: pipe it through to a (possibly closed) special-interest community, e.g. on a LinkedIn group, a Ning site or some such.

[UPDATE, 2011-02-11, 13:47 : For some reason I had overlooked number 12. And number 13 was inspired by Richard Grusin‘s comment below:

12. Link inside a (micro)blog post to stuff that’s relevant to the topic at hand. In fact, links could well be the most important gestures that we should measure.

13. Approve/reject incoming blog comments or track backs. With this one, the negative signal of rejection might be the more significant one.]

Anything else?

These gestures inform the public or people in my on-line communities and on the Internet in general, as to which content gets through my personal cognitive filters, my interest filters, and therefore get amplified and possibly more widely distributed.

These gestures can be used as input for social recommendations. And that includes news. Why not? Actually, the concept of news is in itself quite fluid. A colleague of mine a couple of years back would define news as “something that is new to someone” – information which is new to someone. Looking at it that way, a lot of information can be news.

Booting up a personal recommendation system for news

As I mentioned yesterday, I’m a big fan of ‘Rebooting the News’. That goes for both meanings: I love the podcast series by Jay Rosen and Dave Winer; and I’m also totally intrigued by the phenomenal transition of our system of news which is happening right under our noses.

In the 9-minute passage of RBTN 82 that I transcribed, our hosts talk about an idea that Dave put forward in a recent blog post, ‘Find me stuff that I’m interested in‘. It’s a discussion about the concepts of a personal recommendation system for news, on Dave’s part inspired by collaborative filtering technology which underpins Amazon’s personal product recommendations.

Not only do I agree with all the conceptual choices that Jay and Dave favor, – such as avoiding categories, using gestures, using feeds, looking at other users’ previous behavior, including information about authoring as well as consumption, including serendipity… – ; I have actually been thinking about these exact concepts for years.

Now, I’m not going to say, “It’s all been done already”, because Dave would think I’m trying to pitch a product 🙂  Truth is, had it been done, we would all be using it. A personal system of highly relevant information is pretty much the Holy Grail of the Internet.

One potential complication with applying collaborative filtering to news content is that, when news breaks, there is no critical mass of gestures from previous users. This may cause some delay in the build-up of a recommendation. Instead of immediate, mass-scale amplification of the breaking news event, the news item might be a more slowly developing “trending topic” as per Twitter.

Also, when the news is very fresh, and its relevance is very personal (i.e. highly relevant to a small number of people), it may take too much time for a collaborative filtering system á la Amazon to collect sufficient gestures from other users in order to deliver the recommendation to the right people.

Therefore, rather than waiting for a new news item to pick up the critical mass which can enable collaborative filtering the Amazon way, we could instead look at the *history* of users’ gestures. If the stuff I have “gestured” in the past is very similar to the stuff you have “gestured” in the past, there is a likelihood that what you “gesture” next will be of interest to me.

So what I propose, instead of collecting many gestures from different users in order to generate a recommendation to one specific user, is to identify pairs of users whose gesture behavior is most similar, and let their behavior inform their mutual recommendations.

One could calculate a “similarity-percentage” for each combination of two users based on their gestures. With a view to serendipity, the ideal similarity is not necessarily approaching 100 percent. The system could offer users a feature to mix their own doses of serendipity. Want more off-beat news today? Turn the potmeter down to 70 percent signal and get 30 percent noise!

BTW, one headache which this idea would take care of is the eternal question: “What is news?” Whatever news means to you is defined by what you “gesture”. Hence the more accurate question to ask would be: “What is relevant?” or, indeed: “What is interesting?”

Like said, I’ve been pondering over this stuff for a while and I’d just love the opportunity to help make it happen.