The Granovetter hypothesis assumes that the function of strong and weak ties are quite different in social networks: strong ties usually occur within groups or communities and their role is to hold the given cluster together, whereas weak ties usually appear in between commu- nities, and their role is to hold the different groups (and thereby, the whole system) together.
An interesting related empirical study on a large phone call network is presented in the following paper: https://www.pnas.org/content/pnas/104/18/7332.full.pdf , where it is shown that the critical point of the percolation is at different value of the fraction of the removed links if we remove the links in descending or in increasing order of the link weights. Furthermore, a quantity O measuring the overlap between the neighbourhoods of the link endpoints is also introduced, and again, the disassembly of the giant component into small fragmented components is different if we remove the links according to descending or increasing order given by O.
The main aim of this project is to repeat the link removal experiments shown in the paper on the weighted co-authorship network between scientists that can be downloaded from http://www-personal.umich.edu/~mejn/netdata/netscience.zip.
The notebook includes a full report about the study submitted as part of evaluation criteria for the Network Science subject at the Eötvös Loránd University, Computer Science M.Sc program.