Exploration of existing ML approaches is needed.

Since we would eventually want to know the existing duplicate graph, that could also serve as a middle-ground from where we can continue with ML approaches or simply listing recent duplicate target issues.