You should very quickly talk to a local station and validate that your idea makes sense with regard to business objectives and content restrictions.
Specifically, I think you may find that stations will not let you copy content or ads to clips on your Amazon CDN account as it is not in line with how they manage content rights and restrictions, nor does it allow to them to manage ads and tracking.
I believe you can access segment clips on a local stations website - why not do your alerts/preferences as a marketing feature/service that alerts then direct folks to the clips on the stations' site?
I think this idea is potentially useful and having a news/tv background, I can see the value both the the broadcaster and to the audience. Have you tried offering this to a large broadcaster, such as NBC, as a test pilot, working together to refine the product and use as a case study? NBC is interested and actively integrating emerging digital tech. Local news is a very tough sell as they don't have much money and take limited risk.
i really like the idea. how are you getting around the copyright issues of parsing out their shows and displaying it online?
Clickable Link: http://www.newspvr.tv
Most stations in the US are now posting content transcribed directly to their own websites with clips of the appropriate story already. Promotion is the then directed to twitter/fb on an automatic or selective basis. Is this basically what you are talking about? In the scenario above, people find out about the content through search or directed discovery or one of the aforementioned social streams.
If so, then please understand that I don't think you've thought through your hypothesis and gauged demand enough. News shows purposefully put the best content at the end of a show because they want people to slog through the 15 or so commercials in the 3 breaks before consuming it. Notice this whenever there is a juicy story or big weather prediction.
Enmass show viewership pays more than selectively pulling out the good bits. AftEr the fact, the value of the content would be zero anyway, so it's repurposed for web use.
I would validate your hypothesis with both the consumer and customers. If you can't see it working, pivot. Perhaps a subscribable video playlist based on topics (like good alerts) not channels (like YouTube).
if you use this idea and it generates revenue, I want 2 percent.