How does Watson handle a corpus with conflicting data?
If you uploaded (to the corpus) a document that contained the statement, 'all apples are red' and another document with the statement, 'all apples are green', and you asked Watson the question, 'What color are apples?' - What answer would it give? Would it mention the conflict in it's response?
Accepted answer
The correct answer would be based on training questions you give it to the learn the ground truth.
If you had multiple questions with similar context mapped to both questions, then that is going to cause an issue. If it is a case where the context is different then enough training questions would allow it to answer the questions correctly.