Something that polled trending news in categories like "Entertainment" and pulled key phrases, and then filtered for said phrases might be a good start. Heck,

Something that polled trending news in categories like "Entertainment" and pulled key phrases, and then filtered for said phrases might be a good start. Heck, considering how much of news is just re-posted AP wire content, you could block entire redundant chunks by simply filtering by a string long enough to be a unique phrase (surprisingly short, as little as 6 words [ https://www.plagiarismtoday.com/2012/07/03/how-long-should-a-statistically-improbably-phrase-be/ https://www.plagiarismtoday.com/2012/07/03/how-long-should-a... ] )