3 ms·
I'm assuming that your spam analysis algorithm will result in a "yes/no" answer that is also encrypted. You can return this value to the originating party, havi
by jerf 17y ago
I'm assuming that your spam analysis algorithm will result in a "yes/no" answer that is also encrypted. You can return this value to the originating party, having spent your cycles on computing the answer, but you still don't know what the answer is, only the decrypter does.
That said, I'm still not sure how well this could possible seal off the computation from the data. Even if I encrypt a "yes/no" answer, can't I still return either an empty string for "spam", and a long, long string for "not spam". I very much look forward to a decent description of the actual process from someone who has chewed on the original paper for a while, and can process it down from "PhD in encryption" to "bachelors in comp sci + personal study" with minimal fidelity loss.