One of the problems of analysing social networks is that researchers usually don't have methods which:
- Show how the network formed or changed over time;
- And represent the true physical configuration of ties.
One can easily have one of the two points above separately, but it is quite difficult to obtain them simultaneously. For example, a social network generated by Facebook or Instagram can show how the network changed over time, but cannot say anything about its true configuration in real life. On the other hand, a social network obtained by mining data from publicly available sources or by gathering data through surveillance during an investigation could be a reliable source of the actual configuration but it is a snapshot of the past. That is, once the network has been constructed, some time has passed and the network probably has changed before you can see it.
Our work, which is still in the early stages, starts from the above assumptions and tries to solve this problem by first mining data from the one obtained through our system (that is a platform where students can book a seat in the libraries of our university), generating a social network, and then analyse it.
In this repository there are 3 files:
-
TIRESIA_report.pdf: contains the complete report of the project (read this if you're interested in our methodology and results).
-
TIRESIA_agglomerative.ipynb: contains the code about the Agglomerative Clustering approach.
-
anon_booking_history.csv: it's the dataset we collected and used. The real students' codes have been hidden to preserve their privacy.
The notebook with all the code about our first attempt can be opened in Jupyter or Google Colab.
If you want the code about our second attempt, with our custom algorithm, please contact me at edo.gab33@gmail.com.
The work is made by:
-
Teodoro Caputo: teocap on GithHub https://github.com/teocap
-
Edoardo Gabrielli: edogab33 on GitHub https://github.com/edogab33
-
Luca Palluzzi: randomlulli on GitHub https://github.com/randomlulli