Ask HN: What product can be built around automatic summarization algorithms?

5 points by MojoJolo ↗ HN
Hi, I have an automatic summarization algorithm for web articles (extraction only). What product can be built from it? Or how can it be used as a startup product?

20 comments

[ 3.2 ms ] story [ 51.8 ms ] thread
Create an API and sell access to it. Later if you come up with an idea, you can still build it.
Who can be the target market for those APIs that I will sell?
How much per call, and how long is the output? Example of quality?
Output is 5 sentences. Quality is still being tested but here's a sample:

from this article http://techcrunch.com/2012/09/09/do-you-need-to-be-a-jerk-to... Do You Need To Be A Jerk To Be A Successful Entrepreneur?

- I recently read Ben Austen’s WIRED article about Steve Jobs, which prompted me to put together my thoughts about the tradeoffs of being a successful entrepreneur.

- After reading it, you might be convinced that you can either be a jerk and successful or decent and mediocre.

- Successful people come in all forms and they all have different limitations, baggage, prejudices, and ways of looking at the world.

- Jobs was successful because his unreasonably high goals, brilliant insight, and relentless passion made people want to work with him.

- You don’t have to have a surly personality to be successful, although sometimes it comes with the territory.

App developers.
For what kind of apps I mean?
I would be interested. Contact me.
How about summarizing blog posts for twitter?
I don't think that has been done - that might be worth exploring and would definitely be cool...
This is a cool idea. But the algorithm only extracts the most important sentences. What will happen is it will just tweet the most important sentence in the blog post (considering that it's 140 characters or less).
Is it multidocument summarization? I wrote a term paper on this back in school some years ago and though the technology is very interesting, it left a little to be desired (so I wonder what the state of the art is now). The applications I studied for this are removing bias in e.g. news articles, critic's reviews, etc.
Mine is only single document summarization. Yes, it's an interesting technology but I think there is not much of improvement.
Summly (http://thenextweb.com/apps/2011/12/17/meet-the-16-year-old-w...) does exactly this... It was created by a 16-year-old and generated tons of press, getting VC funding and all...

It's worth noting that the app website is down and it's not available on the App Store anymore; however, according to their Twitter feed, a new version is supposed to be released soon...

Tried Summly, as a matter of fact, it is included in my related systems. I don't know what happen to them that they don't exist as of today. Well, in my opinion, it doesn't summarize that good.

You also might want to check Cruxbot (http://www.cruxbot.com/).

Summly seems to have been redesigned (http://summly.com)... It looks pretty cool, at least from their website and a few sample summaries...
Maybe, summarize the text of emails for people who receive lots of email each day.
You could build a studying / note-taking app. Feed in source articles (textbook, PDF, et all) and have it automatically generate outlines. Customizable SparkNotes. This could be useful to every student in the world.
Can extracted sentences be considered as outline?
The summary for the TechCrunch article you posted below reads to me like an outline. As long as the extractions encapsulate the main themes and ideas of a certain piece, they would be viable study tools.
Have you benchmarked it on some standard IR collections (TREC-DUC comes to mind)?