adding http://tldr.it as a generic approach.
Here's why: Let's assume we have the perfect algorithm that knows exactly which sentences to pick for a good summary. You'll still end up with a summary that's horribly out of context and difficult to read.
Try it yourself. Assume your brain is the perfect algorithm and pick the most important sentences for a summary. Then try to read just the sentences that you picked without rewriting it into a coherent paragraph. More than likely, it'll be an out of context confusing block of text.
Algorithms will never solve the summarizing process unless they are teamed up with a rewriting engine that could build coherent paragraphs.
> Algorithms will never solve the summarizing process unless they are teamed up with a rewriting engine that could build coherent paragraphs.
That doesn't sound like never. Frankly, I wouldn't be surprised if such AI would created within 50-100 years, ie in my lifetime. I think this comment exposes the common false thought-pattern that if something is not viable in the next quarter, then it'll never happen.
I know if I wrote a summary of a Microsoft Surface article (I have one), it'd be very different from someone who loves it -- or is even an MS employee.
And spam... big issue there.
(disclaimer - I have your plugin and have tried it, summaries seem way too short.)