A media monitoring company wants to automatically scan thousands of news articles and pull out every company name, person, and geographic location mentioned in each article so analysts can build relationship maps. Which text-analysis technique should they use?