If you’re like most people these days, you simply don’t have a lot of time to scan entire articles to see if they actually have the information you’re looking for. WikiSeer wants to do the work for you.

“When you browse or search on the Net, you often come across so many different kinds of documents and you don’t know what they’re about,” says the co-founder of the new service, still in beta, Sameer Yami. “There’s so much information overload, and we want to reduce that so you can just read what you are interested in.”

How Wikiseer attempts to reduce the amount of time that you spend searching through documents and more time reading what you’re really interested in is by providing “keynotes” that basically reduce an article to its salient points so that you get an overview of its content before you dive in fully. You can use it for sussing out entries that pop up on popular search engines, directly on web sites like Time or The Economist, or whatever other URL address you type in. When you bring your mouse to the link, its HTML or PDF or Microsoft .doc data trots off to the WikiSeer service, is processed in real time and sent back with the keynote information. The size of an article is reduced by a percentage – say, 60 or 88 percent in under a second for all but the longest pieces – which may be enough to give you the information you need, or at least enough to convince you that you want to read the full-length piece. Large documents, Yami says, might take closer to 15 or 20 seconds.

In order to analyze documents so quickly, of course, a lot of undercover work was involved. Natural language processing on the server side was used to create context (and to continually update that context) by analyzing millions of documents, providing the foundation from which the service can draw out its real-time summaries. It has applied linguistic theories to average out what tends to define the sentences generally considered to be important within documents – the who, what, when, where, and why, if you would. “Prior to your clicking on a link,” Yami explains, “we have created context, and from that context, which is pretty good at understanding documents, documents, based on that context, can be analyzed instantaneously.”

The service plans to add other document formats in addition to those mentioned above, but good keynotes generally currently cannot be delivered for non-English documents (though the technology is scalable to different languages, Yami says), or those where there’s not much language flow to analyze – such as video or images where just a few comments define the topic, or documents where the text is already short or is mostly programming code. Delivering keynotes from the dark or deep web is also possible in some instances, if the user has a log-in to the service and the service permits access via WikiSeer. Yami says they’ll look to partner with some services, like the IEEE, that currently aren’t supporting such access.

The service is just a month old but Yami says plans are to continue improving it to cover as many different kinds of content as possible, as well as make sure it processes larger documents even more quickly. The six-person company is starting to look for additional funding sources, Yami says.

• Don’t forget to propose your startup for our Semantic Web Impact Awards. The deadline is Sept. 15.

Semantic Tech & Business Conference Returns to San Francisco

Semantic Tech & Business Conference returns to San Francisco in June! Join us from June 3-7 for complete coverage of Big Data, Linked Data, Extreme Information Management, and Semantic Web. From breakthrough approaches to solving business problems to the big data implications of fast–evolving technologies, SemTechBiz provides you with an unparalleled interactive experience and delivers tangible business value. We're offering a special early rate when you register by February 17. Sign up now!