I’m excited to announce that this year we’ve launched the ICWSM Data Sharing initiative. Starting this year and moving forward, authors of papers published at ICWSM are encouraged to share new datasets. ICWSM is providing a hosting service for all these datasets where they can be freely downloaded and used by the research community.

This year 11 authors contributed their datasets, with 3 more who approached me during the conference. We were delighted to award the newly created Dataset Award to the Luca Maria Aiello, Martina Deplano, Rossano Schifanella, and Giancarlo Ruffoauthors. They authored the paper entitled People Are Strange When You’re a Stranger: Impact and Influence of Bots on Social Networks and assembled two impressive and large maps of the Anobii social graph. Congratulations!

Thanks to all the contributing authors for their participation and the ICWSM organizing committee for their support. The initiative is off to a fantastic start!

From a practical perspective, clusters are expensive, their administration requires non-trivial time and technical knowledge, and the tools for using them aren’t user friendly. On top of this, algorithms that run on network data are notoriously difficult to parallelize. As a result, few common network analysis algorithms have good parallelized versions or are implemented in standard tools.

Despite all this, the average size of network-based datasets is growing. This means that more and more researchers need to do large network analysis.

Earlier today at the International Conference for Weblogs and Social Media, I ran a tutorial that covered a range of easy techniques for making analysis of massive network datasets possible on standard-issue desktop and laptop computers.

The presentation slides as well as demo scripts are all online and publicly available, so please feel free to take a look.

One participant suggested that I should start blogging about these and other techniques for large-scale data analysis, which sounded like a great idea to me. If you’re interested, please visit this space in the coming days and weeks as I’ll start making regular posts about these and other issues in social data analysis. And if you have questions about or requests that I discuss certain topics, don’t hesitate to email me.

Excited about the New iPad? We are too – and we’re not alone. The Apple announcement generated a lot of buzz on Twitter. Here’s our take on what everybody was talking about: The New iPad in Tweets.

To our surprise, of all the New iPad’s features, 4G LTE took the prize as the most popular one. And while Steve Jobs may be gone, he’s not forgotten – tallying up more mentions than all of the presenters except for Tim Cook.

If you want to know more about how we made the infographic below, contact us.

This was the second Python workshop organized by members of the Network Dynamics group. Plans are already underway for holding another in November. Materials on the first and second McGill Python Workshops are available online.

Please email us if you’d like to be notified about upcoming workshops!

Our department is home to professors with expertise ranging from robotics to computational genetics to artificial intelligence. Our undergraduate and graduate programs draw top students from around the world. It’s a privilege to be a part of such an outstanding community.

The event was co-organized by Professors Jerome Waldispuhl (McGill), Michael Brudno (University of Toronto), and Derek Ruths (McGill). Details including the attendees and topics discussed are available online. The one day meeting concluded on Saturday morning with a friendly, but competitive game of soccer.

In their paper, entitled “The Effect of Mobile Platforms on Twitter Content Generation”, Perreault and Ruths quantify a variety of ways in which mobile Twitters users differ from their desktop counterparts: mobile users tend to engage in conversations, often sharing much more personal information with their followers. Furthermore, they show that the Twitter community is divided into exclusively mobile and exclusively non-mobile users – with few individuals who regularly use both mobile and non-mobile clients.

The results, while general in nature, shed light on the way in which mobile platforms fundamentally alter the way that individuals interact with online social data. With more and more interactions being conducted through smartphones and tablets, we expect this work to predict the way that the majority of people will come to use online communities.

The purpose of this first Python workshop was to introduce attendees to the simplicity of working with the language. Professor Ruths started the session by talking about some of the reasons people are choosing Python over other languages: easiness in learning the language, the amazing power it has under the hood and the speed with which one can go from idea to prototype. He then introduced the containers that are built in to the language (list, dict, set, tuple) and presented ways that code can be made simpler and more readable using standard methods on these containers.

Mathieu Perreault then built, in front of the audience, an application that interfaced with the Twitter API. This allowed him to introduce two powerful modules that are bundled with Python: urllib and json. Attendees were guided through the steps of working with a web API (from reading the documentation to making a GET request) and shown how Python makes it possible to go from an idea to a working Twitter program in less than 30 minutes.

There are plans to organize another workshop later this year once the fall semester is underway. Materials from this last event can be found on the webpage for the workshop.

In my talk, I discussed what spoons, cancer cells, Ancient Roman roads, and online social networks all share in common: their structure tells us a lot about their function. Pulling together a number of my lab’s research topics, I showed how we can make surprisingly good predictions about how cells, societies, and economies will behave even when we have relatively little knowledge about how they work. Knowing only the composition (the parts that compose a system) and the structure (the way those parts are connected), we can learn a lot about a complex system.

On Saturday, I discussed how, using only structure, we can predict the effect of drugs on cancer cells (via biochemical network structure), discover many of the political, economic, and cultural entities that operated within the Ancient Roman Empire (via road structure), and learn how people form relationships (via online social network structure).

For those of you who weren’t able to attend or watch the event online, videos should be up soon. Keep checking back as I’ll post a link to my talk once it’s up.

Update: You can now watch the talk online.