Report on the visit of Dr. Marija Mitrović to Institute Jožef Stefan, Ljubljana, Slovenia, on May 12-23, 2014 based on grant STSM-TD1210-16719
Title of STSM: Towards Understanding Popularity of Science Topics: Analysis of Data from Knowledge Dissemination Portals by Bipartite Networks Modeling and Simulations
It was very nice to come back to Ljubljana and Institute Jožef Stefan after more than two years. I would like to thank to my host, Bosiljka Tadić, and the Institute for the hospitality extended to me during my visit to Ljubljana. I had a wonderful time as a result of their excellent arrangements. Also, I would like to thank to Bosiljka Tadić for devoting her time to our very frequent, interesting and helpful discussions.
The main purpose of this STSM was to start with systematic analysis of processes of knowledge dissemination by mapping the empirical data from Q&A sites onto bipartite graphs and considering the self-organized dynamics of these graphs. As it was proposed, we downloaded the data for 121 Q&A sites of StackExchange network. My initial plan was to spend some time during the visit on developing python scripts for and downloading the data from the websites. Luckily, StackExchange community offers the data freely to community for the research purposes, so I was able to concentrate on the data analysis. Since the datasets for each of the Q&A sites contains a large amount of high-temporal data, we’ve decided, for the beginning, to study processes on one of these sites, Mathematics, where people interested in math ask and seek the best answers to questions trough interaction with each other. I mapped the data onto a weighted bipartite graph, with users as one and questions, as another partition and examined its topological properties in order to understand the structure of connections and how they influence the formation of user communities in relation to different topics. Together with Bosiljka, we also analyzed different types of temporal patterns that are result of collective search for knowledge.
The further steps in data analysis will be to apply the same methodology to other Q&A websites from
StackExchange network and compare the result in order to find universal features which are independent of specific field of knowledge. The knowledge obtained from the empirical analysis will help us to design discrete model of users interaction over different topics. This model will enable us to investigate the self-organizing dynamics on Q&A sites into more details. My host and the Department of Theoretical Physics, Institute Jožef Stefan, were kind to invite me to visit them again in June, when I will continue the work started during this STSM.
I would like to thank to COST Action KnowEscape and to our STSM coordinator Andrea Scharnhorst
for giving me the opportunity to perform this STSM mission and start a collaboration on this new topic for me.
Belgrade Marija Mitrović