You could do a hash of each Q/A pair into a dictionary as you iterate through them and only keep each one if its hash doesn't exist yet. If you're looking to do a fuzzier search, you could do cosine similarity and throw out anything that's got too close of a nearest neighbor.
You could do a hash of each Q/A pair into a dictionary as you iterate through them and only keep each one if its hash doesn't exist yet. If you're looking to do a fuzzier search, you could do cosine similarity and throw out anything that's got too close of a nearest neighbor.