If you are on Windows, you can download the executable here. If you prefer, you can download the source code here.

RECAST is a single-parameter classifier that reads a trace file containing encounters (or communications, interactions etc) among people and labels their relationships.  In summary, if Smith and Johnson had an encounter interaction in your trace file, RECAST can label this relationship as:

Class 1 - Friends: Smith and Johnson see each other regularly and have a high number of common friends, i.e., they are part of the same community.

Class 2 - Bridges: Smith and Johnson see each other regularly but they do not have many common friends, i.e., they are part of different communities.

Class 3 - Acquaintances: Smith and Johnson rarely see each other but they have a high number of common friends, i.e., they are part of the same community.

Class 4 - Random: Smith and Johnson rarely see each other and do not have common friends, i.e., they encounter each other in a random fashion.

Thus, if after a week of encounters I have the following graph G(V,E):
V = the set of persons in the scenario (e.g. Smith)
E = the set of pairs of people who encountered in the past (e.g. Smith encountered Johnson at least once in the past week)


Thus, the execution of RECAST will basically separate the social relationships (regular and intra-community encounters) from random ones:



The colors represent RECAST classes.

Reference: RECAST: Telling Apart Social and Random Relationships in Dynamic Networks. P O S Vaz de Melo, A C Viana, M Fiore, K Jaffrès-Runser, Le F Mouël, A A F Loureiro. Proceedings of the 16th ACM International Conference on Modeling, Analysis and Simulation of Wireless and Mobile Systems (MSWiM'2013). DOI:10.1145/2507924.2507950