Jennifer Zaino
SemanticWeb.com Contributor

DailyMe, the personalized news web site and delivery service, launched a significant update this week, powered in large part by OpenCalais' semantic web technology. It is rolling out technology it calls Newstogram, which enables DailyMe to support dynamic personalization of content, based on what a user actually has been reading rather than requiring them to explicitly state what they are interested in reading about.

Tracking every user a story reads on the site and matching it up with metadata extracted using OpenCalais, as well as some other tools, that help categorize what a story is about come together to help DailyMe get a detailed picture of what each user is interested in and then deliver to those specifications. Clicking on My Newstogram, will let a user will see the categories, topics, people, companies and entities they've been reading about.

Automating personalization based on an up-to-date understanding of what a reader has been interested in has a lot of benefits compared to readers only being able to manually state their interests by checking things off a list. For one thing, they don't always check everything they're interested in because it's time-consuming. For another, interests change and evolve over time, and this feature enables the presentation of personalized news to evolve with them.

As an interest of a reader trends lower on the Newstogram, fewer stories will be presented to the reader on that topic -- unless that person has specified in his profile that that category is of interest. But DailyMe expects most readers to use the implicit route to personalization for the time-savings it enables.

"We use OpenCalais for the biggest piece of the semantic processing," says Neil Budde, President and Chief Product Officer for DailyMe. DailyMe licenses content from 500 news sources, some of which may provide their own metadata, and also layers its own intelligence processing and editorial oversight on top of the data it gets from OpenCalais.

DailyMe has been using OpenCalais even prior to the upgrade, to get much deeper levels of categorization. "With semantic technology from OpenCalais we can go much deeper, be much richer and much more narrow in the types of categories and ways people can pick the news they are interested in," says Budde. "Now, to be able to do implicit personalization, you need that richer understanding of content than you can get from the high level categorization we were getting before using OpenCalais."
Personalization is the future, but it's blended with editorial control over surfacing news, he says. For example, you may not have indicated that you were interested in politics or ever read a single story about them on the site, but you'd want to know the minute big news broke in the area -- for example, U.S. presidential election results.

"Personalization is a great complement in that there is so much valuable useful information and news out there that it's very difficult to find that within any given web site," Budde says. You can search if you know what you want upfront, or you can rely on online publication's editorial processes to create a front page that appeals to the widest possible audience -- which becomes less viable as an exclusive approach as screen real estate shrinks, from web windows to e-readers to iPhones.

"How do you find through these small windows that which is important and necessary to know -- the editorial view -- and that which is of interest to me?" he notes. Personalization helps here in getting both the main stories that people should know and those that are of interest to a particular individual "above the fold," so that a third of screen real estate isn't wasted on something that is neither major breaking news nor something a reader has professed to be of interest to him. That helps to bring readers back to a site more often, because they feel better connected to its content.

Currently DailyMe also offers PDF newsletters with the full text of all stories that match a user's interest profile, and four Kindle offerings. For example, the DailyMe Literati for the Kindle is a full-text collection of information about the book industry and book reviews drawn from its 500 sources for which readers pay a subscription fee. Primarily the site works on an ad-driven model, which can be enhanced by using its semantic knowledge of what users are interested in and showing them the ads most relevant to that.

But it's also interested in using its Newstogram technology to help other news publications online improve their content as well, with DailyMe serving in large part as a showcase of what is possible. By delivering to clients a picture of what their users are most interested in, it can help content publishers both better personalize the information they deliver and understand at a very detailed level what their readers are most interested in -- not just the ten most popular stories, for example, but the ten most popular people being read about. That can help outlets tailor their content to those needs. It hasn't signed any deals in this area yet but, says Budde, stay tuned.