HACKER Q&A
📣 tareefdev

How to programmatically recognize main news of the week?


I like to read news on weekly basis and there are so few weekly Arab newspapers. I am thinking about writing a scraper to fetch news on daily basis from several sources, rank them somehow (how?) and create me a newsletter.

How do you may approach the ranking stage? Based on what I can filter fetched articles to keep only the most important ones.


  👤 aguaviva Accepted Answer ✓
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.