Financial text often covers several companies at once. How would you find the main companies, extract the passages about each, and strip boilerplate?

Financial text such as news and research reports often covers several companies at once. How would you (1) identify the companies that are the main subject of a text, (2) extract, for each of them, the passages that discuss it, and (3) strip out non-essential material such as boilerplate and disclaimers? Describe the method, the implementation and the compute it needs, compare alternative approaches, and name the main difficulties and bottlenecks.