Preferential attachment: be first

Preferential attachment is a key process governing the dynamics of many economic, social and biological process. It is the “The rich get richer” mechanism by which a quantity is distributed among individuals according to how much they already have. It also happens in social networks and the ones that have more social connectivity (the “hubs”) receive more new connections than the poorly connected. In a famous paper, Laszlo Barabási and Reka Albert encoded this mechanics in the so called Barabasi-Albert model to generate random scale free-networks. ...

#R #Temporal Networks #igraph #Twitter #Social Networks

Temporal network of information diffusion in Twitter

Millions of tweets, retweets and mentions are exchanged in Twitter everyday about very different subjects, events, opinions, etc. While aggregating this data over a time window might help to understand some properties of those processes in online social networks, the speed of information diffusion around particular time-bound events requires a temporal analysis of them. To show that (and with the help of the Text & Opinion Mining Group at IIC) we collected all tweets (750k) of the vibrant conversation around the disputed subject of the general strike of March 29th in Spain. ...

Algorithms and Management

Yesterday I gave a talk in the 6th IIC Technology Conference about how Social Contagion can be leveraged for marketing purposes. The motto of the conference was about the need of using Algorithms in nowadays business process. With the availability of more and more complex data the use of algorithms that can detect and reduce complexity is of paramount importance. Big data is not only about volume (TeraBytes of data), it is about huge complex data and reducing that complexity can only be achieved by modeling, simulating and analyzing the patterns we observe in the data. ...

#Social Networks #Recommendation algorithms #Stochastic Blocks

Predicting Human Preferences Using the Block Structure of Complex Social Networks

Authors: Roger Guimerà, Alejandro Llorente, Esteban Moro y Marta Sales-Pardo Journal: PLoS ONE 7, e44620 (2012) LINK Abstract: With ever-increasing available data, predicting individuals’ preferences and helping them locate the most relevant information has become a pressing need. Understanding and predicting preferences is also important from a fundamental point of view, as part of what has been called a “new” computational social science. Here, we propose a novel approach based on stochastic block models, which have been developed by sociologists as plausible models of complex networks of social interactions. ...

It was a delicious assignment, infinitely complicated

I have just read an amazing book “Shibumi” by Trevanian (a.k.a. Rodney William Whitaker) probably the best spy novel I have read so far. In the book, a big data computer (called Fat Boy) is operated by a “data scientist” (although is not called that way). I enjoyed very much the following paragraph, an analogy of the emptiness of big data without insight and also a musing about how difficult is to find relationships from activity data (the kind of research we do! ...

Which chart to use

In most of my talks I present quantitative evidence of patterns, data exploration or results. But which is the right way to show that evidence? Worry no more: the Extreme Presentation Method helps you to decide with this chart chooser (click here to download the pdf) My impression is that this chart chooser is good for small data. For big data some of the charts render useless. For example, read the insightfull post Don’t use Scatterplots by Chris Stucchio on why is not a good idea to use scatterplots to show relationship between bivariate data when you have large amounts of data. ...