This is the first implementation of the COnfECt method, used for preliminary tests.

COnfECt is a model inference method that infers LTSs (Labelled Transition Systems) for each component of the system from execution traces. It is based on kTail, a passive model learning method that works in two steps:

• First, it constructs a tree from the traces, with each branch of the tree corresponding to a trace, and each event to an edge.
• Then, it merges all states of this tree that have the same k-future, i.e. the states that have the same future of length k.

The method COnfECt adds two steps to kTail: Trace Analysis & Extraction, and LTS synchronisation.

In this step, we try to separate events in a trace that are due to different components.

For that we calculate a Correlation Coefficient for each pair of events, and we define a treshold above which two components are due to the same component (we say that they have a strong correlation). This Correlation Coefficient is used to separate each trace into many traces, containing events of only one component.

Example:

Trace		
/devices 
/json.htm(idx:=115,svalue:=15.00) 
Response(status:=200) 
Response(status:=200,data:=[1]) 
/json.htm(idx:=115,svalue:=16.00) 
Response(status:=200) 
/devices 
Response(status:=200,data:=[1]) 
/hardware 
Response(status:=200,data:=[2]) 
/config 
/json.htm(idx:=0,switchcmd:=On) 
Response(status:=200) 
Response(status:=200,data:=[2]) 
/tools 
Response(status:=200,data:=[3])

STraces = {
T1 {
/devices 
call_C2 return_C2 
Response(status:=200,data:=[1]) 
call_C3 return_C3 
/devices 
Response(status:=200,data:=[1]) 
/hardware 
Response(status:=200,data:=[2]) 
/config 
call_C4 return_C4 
Response(status:=200,data:=[2]) 
/tools 
Response(status:=200,data:=[3])
}

T2 {call_C2 
/json.htm(idx:=115,svalue:=15.00) 
Response(status:=200) 
return_C2}

T3 {call_C3 
/json.htm(idx:=115,svalue:=16.00) 
Response(status:=200) 
return_C3}

T4 {call_C4 
/json.htm(idx:=0,switchcmd:=On) 
Response(status:=200) 
return_C4}}

Each trace obtained with this step will produce a LTS with the first step of KTails.

The goal of this step is to join LTSs that contain similar events, and model behaviours of the same component.

Three strategies are implemented:

• the Strict strategy : we want limit the over-generalisation. We do not join LTSs, and we cannot repetitively call an other component.

• the Weak strategy : we want to reduce the number of components. We join LTSs, and allow repetitive call of an other component.

• the Strong strategy : we obtained a more general model. We join LTSs, and each LTS can call any other LTS anytime.

Then the models are generalised with the second step of kTail, to obtain the final LTSs.

traces/ contains the different test cases (Exp1-Exp7, and test cases for the time execution).

results/ contains the results of the different experiments.

COnfECt/ contains the implementation of the COnfECt method.

Put all your traces in the folder COnfECt/traces. If the folder COnfECt/COnfECt does not exist, make it. It has to be empty.

Go into the COnfECt/ folder and execute the script exec.sh with the synchronisation strategy as arguments (strict, weak, or strong):

cd COnfECt
./exec.sh strict

Results are generated in the COnfECt/COnfECt folder.

• put the correlation coefficient and similarity coefficient as arguments