If you had a button that you could press to pause time, make flying birds freeze in mid air, etc. what would you do with that opportunity? Some of us would catch up on all the links we’ve bookmarked as “to read” or favorited on Twitter. I don’t have a button like that, but a new iPhone app that launched today comes pretty close.
Summly (iTunes) is a machine learning app that uses semantic analysis of text and a variety of algorithms to cut long-form text down into just a few hundred characters; it summarizes articles online by capturing just a few of the most important sentences. It does a good job and is available for free in the iTunes store. It’s a powerful tool that I’ve been anxiously waiting for ever since its founder, 16 year old UK dweller Nick D’Aloisio, announced that he had raised venture capital and was shutting down his first iteration of the app, called Trimit. Summly is the new version of the app, it’s free and it’s really easy to use.
Founder D’Aloisio today launches the iPhone app and API, he says a web version and new features like automatic import of Twitter favorites will come soon. The app includes a bookmarklet you can install into mobile Safari and click on any webpage to launch Summly’s automated summarization app.
I like it quite a bit and am very glad to see the app relaunched. D’Aloisio has raised money from Horizon Ventures, the investment arm of Chinese multi-billionaire Li Ka-shing.
Robot, Please Read For Me
D’Aloisio won’t go into great detail about his patent-pending summarization technology but says that Summly first identifies the general category an article is in. He’s got one set of rules for summarizing articles about politics and another for summarizing articles about sports, for example. The technology takes into account keyword frequency, many other factors and learns over time.
Machine Learning expert, data scientist and summarization technology lover Hilary Mason says that there have been simple technologies to do things like this for some time, but that very few compelling products have been made out of them for end users. Mason was particularly pleased with Summly’s search capabilities, which D’Aloisio says are based on a number of 3rd party search APIs, some of which are piped directly into the summarization technology and some of which aren’t.
D’Aloisio says his technology is far from simple; Mason says that’s just how she would go about it. Time will tell whether simplicity or complexity is most effective, but either strategy will go on within the context of a mobile user experience that in the case of Summly seems pretty good.
The end result is really easy to use and is a big time saver.
It goes without saying that there is more good information available online today than people can manageably absorb. Automated summarization algorithms are a great way to catch some highlights and get a sense if you want to dive all the way in to read a full article. In many cases, though, I’m satisfied with Summly’s short summaries. Some summaries are better than others and I’d love to know more about other peoples’ experiences with the app and thoughts on the quality of its results. There are some little changes to the analysis and the user experience that I think would make a big difference.
On balance I’m really happy to have access to the app though. It could be better and hopefully it will be in the future – but I don’t have anything else like this in my life. And I have a whole lot of things I want to read. I bet you do, too.
