Local discovery app WHERE has partnered with mobile startup Bump for a new local recommendations service called Perfect Places. Bump’s technology, which lets developers build apps that share data when users tap their phones phones together, is already available in a number of applications including PayPal, HootSuite and a universal remote control. With the new WHERE integration, mobile users can now bump phones to share local recommendations with each other. The app matches up the relevant similarities in users’ interests for its suggestions, and it works even if the two end users are not yet friends in WHERE.
WHERE CEO Walt Doyle says his company is on its way to owning the “pre-checkin space,” referring to the location-based social networks like Foursquare and Gowalla which let you register your location at a particular venue via smartphone applications, aka “check in.” Instead, WHERE wants to focus on the problem that arises before the check-in: figuring out where to go.
Users of the WHERE app simply launch the application on their device and then tap their phones together in order to share local recommendations. Of course, this assumes that each person has already built up a taste graph of places they like.
WHERE’s PlaceGraph Explained
This taste graph – or place graph, rather – forms the basis of WHERE’s recommendation engine. It’s built up as you interact with places – by saving them in WHERE, reviewing them, going there and checking in (the app lets you check in via Facebook Places), liking them and more. But even if you’ve only casually used WHERE in the past and have never taken any of these direct actions within the app, it still may know something of your interests.
That’s because the place graph uses both explicit actions (ratings, reviews, check-ins) and implicit actions (browsing a particular a particular category or viewing the place detail page for a given location) in combination with place info data to build your customized taste graph. Obviously, the explicit actions are given more weight in WHERE’s PlaceGraph Algorithm, the technology that builds this list of your preferences, since implicit actions could be the result of accidental clicks.
The interesting thing about WHERE’s recommendation engine is not only that it works in real-time, but that it’s continually and dynamically updated as you move around town. Like sushi? WHERE doesn’t just recommend sushi restaurants all around town, but will surface just those nearby. However, there’s also a “serendipity” component to WHERE’s technology that provides users with unexpected, personalized local discoveries too. In other words, it’s not as simple as “Like sushi? Here’s more sushi.” But the recommendations here aren’t random – they’re surfaced by computing which nearby nodes are connected to a user’s places profile and are strongly connected to place unrelated to that user. These serendipitous recommendations make up about 20% of WHERE’s suggestions.
It will be interesting to combine this customized taste graph of places with another user’s, but it still feels a bit incomplete without taking into account other checkin-based social networks a user may have accessed in the past. Why not pull in Foursquare and Gowalla checkins, local “likes” from Facebook, or data retrieved from competing services like Yelp or Urban Spoon? Until any given app looks beyond its own borders at all this information combined, it can’t truly know our tastes. It only knows a portion.
(Who’s building such a service right now? If you are, please get in touch.)
WHERE Perfect Places is available on the Android and iPhone versions of the app currently.
