The scraper indexes each paragraph into a separate document. And fetches 3 matches / snippets from each article by default