The Tool
In this page, you can find the instructions to reproduce the experiments performed as part of the paper "A Genetic Algorithm for Goal-Conflict Identification" submitted to ASE 2018.
Installation Instructions
REQUIREMENTS
Java 8.
The tool has been tested in Linux and MAC OS.
DOWNLOAD
You can download the tool from this link: conflict-detection-by-learning.
INSTALL
To compile our tool run: ant compile.
We provide an Ant script (build.xml file) and the libraries required (JGAP and ltl2buchi) into the lib folder.
We provide two different LTL solvers:
aalta is the one used in the experiments, but it only works in Linux OS.
pltl is an LTL Solver that works in Mac OS.
Running the Experiments
In the folder case-studies you can find the scripts with the specifications of each one of the case studies used in the paper.
To run the scripts you have to be in the root folder.
For instance, to run the Rail Road Crossing System case study, first move to the root folder, and then execute ./case-studies/rrcs/rrcs.sh
Notice that, if you want to modify any parameter of our genetic algorithm (e.g., the population size, number of generations, size of the chromosomes, etc.), you have to modify the java source class src/main/BCLearner.java and recompile the project.
Reading the Results
We ran the algorithm 10 times for each case study. You can find the results of all these executions, and the results we obtained by using the tableaux-based technique, in the folder case-studies-results.