You are building a model to improve the targeting of your company's online advertising campaigns. You need to create a training dataset and want to prevent the creation or reinforcement of unfair bias. Which two actions should you take?