Some mix of topic clustering + trending-ness (these being both well-studied topics that one should be able to easily find papers + working examples for) should

Some mix of topic clustering + trending-ness (these being both well-studied topics that one should be able to easily find papers + working examples for) should do the trick. More specifically: first topic-model, then aggregate the trending score over those buckets. Then sit down and enjoy your news + coffee. I'd be curious as to which Arabic-language newspapers you think are especially worth reading, BTW.