Something that disqualifies things that don’t contain keywords from the thing you’re currently working on might function as a very crude version of this.
Come up with 20 independent algorithms like that, use bayes theorem to combine the results, and you should come pretty close.
Simple compared to what, and with what rates of false positives/negatives?
As in implementable in a couple of hundred lines of JavaScript. If I had good answers for the second question, I’d be a lot more sure than 20%.
Something that disqualifies things that don’t contain keywords from the thing you’re currently working on might function as a very crude version of this.
Come up with 20 independent algorithms like that, use bayes theorem to combine the results, and you should come pretty close.