Frequent subgraphs mining for time-dependent graphs

Question

I need to know if there are "frequent subgraph mining" algorithms that take into account time information on edges.

Let me explain,

I have a graph in which, nodes represent building rooms, and the edges represent the movement of an occupant among the rooms. Example: R1 and R2 are two rooms, (R1,R2, T) represents the movement of the occupant between the room R1 and the room R2 in time interval T. For each day, occupant moves within building rooms then these movements are saved in a graph. The same thing for all other days.Example: if we have 7 days, we will have 7 graphs. I want to know if there is "graph mining algorithm" or "graph algorithm" that allow extracting frequent subgraphs in a specified time interval, to learn the frequent movement of the occupant among building rooms.

Remark: I think that Apriori-based approach and Pattern-growth approach do not consider time-dependent graphs.

Thank you for your help.

Answer 1

I know only one graph pattern mining algorithm which takes time into account: GERM (Graph Evolution Rule Miner). There was also a C implementation of this algorithm. It is an adaptation of gSpan: the DFS encoding was re-defined, and a different definition of a match (between a query and data graphs) is used. You could potentially use this algorithm as a helper to design the solution for your use case.