Semi-Targeted Password Cracking via Keywords
We present a novel method for password guessing driven by keywords related to a particular service. By using these keywords, the model is able to make more-informed guesses, requiring fewer guesses than a completely general model. However, unlike other targeted models, this model doesn’t require private or personal information such as a user’s birthday or other leaked passwords. Using a set of keywords related to the service and prior known passwords from similar services, we improve the percentage of guessed passwords by up to 3% without any specific data from a particular user.