Research paper: Personalized news recommendation based on click behavior

Online news reading has become very popular as the web provides access to news articles from millions of sources around the world. A key challenge of news websites is to help users find the articles that are interesting to read.

In this paper, we present our research on developing personalized news recommendation system in Google News. For users who are logged in and have explicitly enabled web history, the recommendation system builds profiles of users’ news interests based on their past click behavior.

To understand how users’ news interests change over time, we first conducted a large-scale analysis of anonymized Google News users click logs. Based on the log analysis, we developed a Bayesian framework for predicting users’ current news interests from the activities of that particular user and the news trends demonstrated in the activity of all users.

We combine the content-based recommendation mechanism which uses learned user profiles with an existing collaborative filtering mechanism to generate personalized news recommendations.

The hybrid recommender system was deployed in Google News. Experiments on the live traffic of Google News website demonstrated that the hybrid method improves the quality of news recommendation and increases traffic to the site.

via Personalized news recommendation based on click behavior.

On we try to emphasize the analysis of authoring behavior over consumption behavior. At this stage we don’t recommend based on content. We do apply collaborative filtering.

(Hat tip to KP)


Open invitation to your link recommendations at (#uutisraivaaja)

This week I’ll be inviting some of my Facebook friends to join I thought I might just as well open this up for everyone to come and try it. So, if you’re reading this, feel free to join the club! 🙂

The invite goes something like this:


I hope you are well.

Over the past few months I’ve been working with some friends and colleagues on a project called ‘Gestures’, or ‘Eleet’ in Finnish.

We are building a website that will offer users recommended links based on the links people share on-line on the Social Web, such as on Facebook.

With this project we are taking part in an innovation competition sponsored by the Helsingin Sanomat Foundation. The contest is called ‘Uutisraivaaja’ (‘Newsplorer’). We’ve made it to the finals and the August 1 deadline for showing what we’ve accomplished so far is approaching fast. We will publish our source code in September.

In a recent release we have implemented Facebook integration. This means that if you log on with your Facebook account, our system will offer you some recommended links based on what you and other people have shared on Facebook.

So, this is my invitation to you to come to and see what you think of it.

It’s very early days and the system is far from perfect. You are actually among the very first people I’m inviting to come and test it. The quality of the recommended links will improve as we get more people to use the service.

Also the user interface may show some glitches as the site is still heavily under construction. Don’t hesitate to use the feedback form, even frequently! 🙂

I look forward to hearing what you think!



P.S.: Feel absolutely free to spread the love 🙂

P.P.S.: Looking at some initial feedback, it appears that when new people log on to , they may not immediately see links recommended to them, but in many cases they do when they come back after an hour or so. Our system needs some time to match you with other similar users, and to collect interesting links to recommend.

But in some cases, depending on the links you’ve shared on Facebook, no links can be rcommended even then. Then it’s just a matter of time when people similar to you will start using Eleet.

Minna Ojamies Pitäjänuutisten päätoimittajaksi | Pitäjänuutiset

“(…) Minna Ojamies Pitäjänuutisten päätoimittajaksi

Paikallislehden pitää olla mukana kaikessa, mitä paikkakunnalla tapahtuu, Minna Ojamies toteaa.Ympyrä sulkeutuu syyskuun alussa, kun Minna Ojamies aloittaa Pitäjänuutisten päätoimittajana. Minna tuli lähes suoraan lukiosta lehden kesätoimittajaksi 1985, kävi Otavan opiston tiedotus- ja viestintälinjan ja palasi Pitäjänuutisten vakituiseksi toimittajaksi jo seuraavana vuonna. (…)”

via Pitäjänuutiset verkossa.

And what do I read in my local news paper this morning? My wife will be its new head editor starting September 1st. Neat! 🙂

3 Reasons Why Relevant Content Matters | Social Media Explorer

“(…) brands need to use paid media (display ads, search, out of home, broadcast), earned media (influencer and advocate outreach programs, events) and owned media (Twitter, Facebook, YouTube, blogs) to reach consumers with the same and or similar messages.

Secondly, their messages have to be relevant. The Edelman Trust Barometer indicates that individuals need to hear/read/see things three to five times before they actually believe it. (…)”

via 3 Reasons Why Relevant Content Matters | Social Media Explorer.

Interesting. So, only “secondly” do messages sent out by brands need to be relevant? 🙂


Nokia’s former employees are building Finland’s future – TNW Europe

“(…) Petra Söderling, Chair of the Board of Directors at Mobile Brain Bank, tells me that there are “tens if not hundreds” of ex-Nokians who have gone on to form their own companies. In particular, from the round of job cuts that took place at Nokia in 2008/2009.

And Hanna Manninen, founder of Finnish PR firm in2PR, directed me towards a plethora of startups with former Nokia employees at the helm, such as Marko Anderson/Futureful, Jos Schuurmans/Cluetail, Kari Laurila/Newelo & Bjong, Heikki Ailinpieti/Saagatec, JP Salmenkaita/Osumus Recommendations, Harri Honko/GreyCrunch, Yasin Hamed/Sfonge, Oliver Bremer/Founder2be and Risto Suoranta/Notava, to name but a few. (…)”

via Nokia’s former employees are building Finland’s future – TNW Europe.

Eli Pariser: Beware online “filter bubbles” | TED talks 2011

So Eli Pariser warns us for the ‘Filter Bubble’. Very interesting in the context of the recommendation engine we are building at

We do subscribe to Clay Shirky‘s point that information overload is not the problem but filter failure is. And we are working to do something about that problem by developing a personalised recommendation engine informed by on-line social gestures.

I do agree with Eli that we should be very transparent about how our filters work, explaining what is being filtered-in and what is being filtered-out, and enable users to manage those filters. Those are valid points.

Our engine puts people at the centre of the filtering mechanism. Our algorithm does not so much look at what you click on and read. Instead of looking at users’ information consumption behavior, we look at their authoring behavior. And we look at the authoring behavior – i.e. the on-line social gestures – of their peers.

That – among other things – makes us different from the examples of Facebook, Google and Yahoo! that Eli mentions.

Hat-tip to Marko.