Algorithm 3.

Frequent Evolving Pattern Mining Algorithm

1:	procedure frequent_evolving_pattern_mining($D{S}^{\mathrm{*}},{min}_{\mathit{freq}},\sigma ,\tau ,\delta$)
2:	Initialize $D{S}^E$ as a data set of evolution networks $ℬ,ℰ,ℳ,C,𝒮,𝒟$ as empty sets for birth, expansion, merge, contraction, split, and death events
3:	for $CIG\in D{S}^{\mathrm{*}}$ do
4:	Initialize $E$ as an empty evolution network
5:	Identify Connected components in $CIG$
6:	Populate $ℬ$ with birth events detected as vertices with in-degree = 0 and all their neighbors in connectivity radius $\tau$
7:	$ℒ$ = Edges ordered based on their starting times in reverse order
8:	Populate $ℳ$ with merge events by identification of edges in the order of $ℒ$ whose removal increases the number of connected components with size $\ge \delta$
9:	$ℱ$ = Edges ordered based on their finish times
10:	Populate $𝒮$ with split events by identification of edges in the order of $ℱ$ whose removal increases the number of connected components with size $\ge \delta$
11:	for $component\in ℳ/𝒮$ do
12:	Populate $ℰ$ = with expansion events in $\mathit{component}$
13:	Populate $C$ = with contraction events in $\mathit{component}$
14:	for $component\notin ℳ/𝒮$ and $component\subseteq CIG$ do
15:	Populate $ℰ$ = with expansion events $\mathit{component}$
16:	Populate $C$ = with contraction events in $\mathit{component}$
17:	Populate $𝒟$ with death events detected as vertices with out-degree = 0 and all their neighbors in connectivity radius $\tau$
18:	Create evolution network $E$ composed of vertices representing by evolution events and edges representing the connections among evolution events based on their component/sub-component relationships
19:	Append $E$ to $D{S}^E$
20:	frequent_evolving_patterns = PATTERN_MINING($D{S}^E,{min}_{\mathit{freq}}$)
21:	Return frequent_evolving_patterns