A new matchmaking app from one of the founders of Adaptive Semantics hit Facebook yesterday. Adaptive Semantics, you may recall, developed the JuLiA semantic text-parsing technology that’s now part of AOL’s toolkit, courtesy of its Huffington Post acquisition.
Kingfish Labs is the startup that created Yoke, and it includes Jeff Revesz as CTO. Rob Fishman, who was Huffington Post's social media editor, is the CEO of the company, which recently received $500,000 in seed funding. Yoke's take on the online dating scene is to bring people together with the help of an ontology graph: Its algorithms explore entities, the connections between them, and the strength of those connections to discover common interests between people that just might lead to a real-world bond.
Yoke is deeply connected into the Facebook API, Revesz says. With users’ permission, it accesses basic data such as birthday, location, and education history, and also pulls their Likes in music, bands, artists, movies, books and some general areas outside those categories. Ditto for their closest friends (again, with respect to their privacy settings, so no guarantee as to how far it can get for each individual). Behind the scenes, Yoke mashes up its Facebook Graph data with data from Amazon, Netflix, and Echonest (which powers Spotify radio) to produce an ontology of interest entities for connecting users together. These three sources were chosen, Revesz says, because they’re the easiest to work with, the biggest and the best.
“We’re looking both for similarity information and ontology information,” he explains – that is, for example, how closely two movies might resemble each other, and what entities they might share in common, such as the same director or actors. So, if someone likes one particular movie, the ontology of interest entities can be used to show other people who like similar things.
The Dolly Levi-types can use Yoke to see how a single friend they’d like to set up appears in context with the matchmaker’s other single friends/potential dates – what may be common interests among the two, such as the discovery that both enjoy Jay-Z, or perhaps that Anne likes Jay-Z and Bob likes a similar artist, such as Kanye West. It also discovers through a Facebook API call mutual friends of the two individuals that aren’t even in the matchmaker’s own circle. “So now we know they like two similar artists and both are friends with [the person trying to set something up], but there could be a friend Diane out there who [the matchmaker is] not Facebook friends with,” Revesz says. That’s another interest link to prick curiosity when Dolly tells Anne about Bob, and Bob about Anne.
“That kind of strengthens the connection. The most interesting aspect of data is these kinds of strange, mysterious connections between people, and it’s interesting to see that data come out,” Revesz says. “That’s never been done before. Even applications like Glancee and Highlight are not doing that kind of similarity information. They are not penetrating deep into the graph, and they are not focused on an application – just on who is around you.”
The service takes the same general approach to ferreting out connections and similarities for users looking to make matches on their own. Yoke just went public yesterday, and already has several thousand signups, he says. The team is busy planning the 2.0 release in May, in time for TechCrunch Disrupt in New York City. Revesz promises a full redesign, a mobile version, and an updated feature-set. While he wouldn’t be specific, the big plans are likely to involve doing more to leverage the Facebook Open Graph protocol.
With the seed funding in hand, the company is focused on growth. It’s not about the money, yet. But monetizing plans will not include selling data or requiring a sign-up fee, Revesz promises. Later down the line, he sees possibilities to pair up with businesses, such as restaurants or concert venues, that could be the perfect date site for a new-made match. Yoke could suggest a particular sponsored option based on what it knows about the couple’s likes.
There’s also the possibility of putting in place some sort of virtual gift options users can buy for credits, and maybe real gifts at some point, too. “Things like if you want to flirt with a mix tape, you can send someone music. We have many ideas and we’ll be open to implementing a few of them once we have a user base,” he says.
Speaking of user base, Revesz is not just an exec at the company – he’s also a client. “I definitely am,” he says. “And it works.”