Early in the summer, The Semantic Web Blog introduced readers to Nara, an advanced neural networking service to automate, personalize and curate web dining experiences for users. (See that story here.)
The service is moving ahead with the launch today of its mobile version, as well as in other respects. “We’re now doing a full-on consumer launch of a polished product on both the web and mobile [platforms],” says CTO Nathan Wilson. “People really are clamoring for the mobile component, especially for this [dining] use case.” Versions for both the iPhone’s iOS and Android operating systems are available.
Simplicity matters to mobility. Nara’s been focusing on making it as easy to onboard people as possible, squeezing the most information out with the fewest number of questions, for instance, he says. One new aspect that helps both the mobile and web experience is how it now furthers connecting to someone’s personality. “Before it was very classically semantic web – you give us examples of restaurants you like and we extrapolate from that,” Wilson says. Now its algorithm provides a higher-level entry perspective that also has people communicate their dining personality by clicking on grids of pictures – types of table settings or dishes that appeal to them, for instance, and even activities they’d spend their days off doing. “We connect that type of information to the classical semantic information,” he says.
The service also is now up to 25 cities from the eight it had earlier in the year. The international component will be represented, too, starting with cities in Canada. One aspect from its algorithm improvements point of view is enhancements to cold starts for users, going beyond the city level. Now it takes what it already knows about a user in, say, Manhattan, and extends that to a specific new neighborhood in a different city to make a dining match. “No matter how little it knows about you in a very far-away, remote category, it can cross-connect and in real time bring information back to you,” Wilson says. “It’s the ability to infer, really,” he says, noting there is “crazy pre-indexing” now of everything to rapidly do comparisons.
Speed and scale go hand-in-hand with what Nara aims to accomplish. It added five additional cities to the initial eight a month after launch, and then 12 more in a week’s time, he says. “So we are growing exponentially because everything’s finally gotten automated,” Wilson says. “That’s handy because now that we can analyze multiple categories, it’s getting easy for us to pull in category after category of data and auto-index and personalize that. This could be the web’s first automated personalization engine that relies on the whole web.”
Wilson says there are a growing number of partners that will have Nara working behind the scenes on their site, with Nara translating what it’s done for restaurants to categories ranging from hotels to wine recommendations to even shopping for individual items. Over the summer it scaled its server tier on the Amazon cloud to have any number of servers needed available to handle API requests, important to its goal of being the web personalization platform provider to multiple clients.
Says Wilson, “it’s exciting to give the gift of personalization to other sites and make them more semantically inclined.”