alessandro.lomi@usi.ch
Dear Vlado (and Nusa) – I hope you are both well and keeping safe.
I am resuming communication as we agreed we would do.
Is the coming week a good time to send you a sample of the data and start exploring possibilities for collaboration?
I note that the National Science Foundations of Switzerland and Slovenia have an agreement to support cross-country projects. It is probably too early to discuss the matter, but if our ideas about temporal networks are convergent perhaps we may consider this possibility.
Looking forward to hearing from you in due course,
Alessandro
Federica,
Would you please share with Vlado and Nusa a subsample of your data. I suggest that you send a sample of data covering one year of transactions before the first financial crisis and one year after it. The data will have the conventional relational event structure < s, r, t, w >. If you have this information easily available the weight (w) could be the value of the transaction in thousands of Euros. Do not hesitate to let me know if anything is unclear, Best, aL
Federica Bianchi federica.bianchi@usi.ch
Dear Professor,
It is a pleasure to share our data with Vlado and Nusa.
I send in attachment transactions in year 2006 and 2009, i.e., the year before the first market turmoil occurred in August 2007 and the year after the collapse of Lehman Brothers in September 2008. I organized the data sets exactly as you suggested.
I am available anytime for sharing other material, if needed.
Kind regards,
Federica
Vlado, Nusa,
Here is a sample of our data – as you requested. The data set that Federica has collected and analyzed in her dissertation is very large and covers more than a decade. This is a small sub sample.
The reason for this specific choice of sample is the presence of an exogenous shock that is very likely to have changed patterns of transactions. Hence, any method designed to detect change over time should be able to detect it in these data (as you might recall the original idea I mentioned was the detection of role change, and this is why I originally thought of blockmodeling (but first we need to agree on what “role” means in this context)
Just to give you a general background, the data are produced by transactions among banks taking place on a European electronic market for liquidity. The data provide a very good example of temporal network – where we do not have “ties” but time stamped events connecting senders and receivers (of funds).
I do not know the NetsJSON format that you mentioned in your prior mail, but the data attached have the conventional structure of relational event data. I am also not sure that the data could provide an appropriate context to test routines in the TQ (Temporal Quantities) library and the Ianus program mentioned in your earlier publications. Obviously, these are issues that we will need to discuss.
Thank you for your interest and your time,
Alessandro