Named entity recognition and part-of-speech tagging
Named entity recognition, or NER, is the task of scanning text and automatically identifying and labeling names of people, organizations, locations, dates, and similar categories. Given 'Maria joined Acme Corp in Berlin last March', an NER system should tag 'Maria' as a person, 'Acme Corp' as an organization, 'Berlin' as a location, and 'last March' as a date. This sounds simple but is genuinely hard: the same word can be a person's name in one sentence and a company name in another, and new entities appear constantly that no system has seen before.
Part-of-speech tagging is a related but different task, labeling each word in a sentence with its grammatical role, noun, verb, adjective, and so on. In 'the dog runs fast', 'dog' gets tagged as a noun, 'runs' as a verb, 'fast' as an adverb. This matters because the same word can play different grammatical roles depending on context, 'runs' is a verb in that sentence but could be a plural noun in 'she went on two runs'. Getting this right underpins other tasks, like figuring out who did what to whom in a sentence.