Led by Ex-Microsofties, Raveable Makes Sense of User Reviews, Gives Hotel Ratings at a Glance
Raveable is a year-old Seattle-area startup that launched its hotel review summarization website in May. If there were a Raveable entry for Raveable itself, here’s what it might say:
Ranked 116 out of 340 tech startup websites in Seattle.
The good: Team is ambitious and knowledgeable; large market; useful technology; fun interface; customer focused; strong word of mouth.
The bad: Relatively new; pre-revenue company.
Best kept secret: Gaining attention from angel investors and VCs.
The idea of Raveable is to help leisure travelers quickly make sense of all the user reviews out there on the Web, and choose a hotel that’s right for them. So the company aggregates reviews from sites like TravelPost (Kayak), MyTravelGuide (Priceline), CitySearch, Yahoo Travel, and VirtualTourist, and provides a bullet-point analysis of the pros and cons of each hotel—for 55,000 establishments and counting in the U.S.
The company was founded by Philip Vaughn and Rafik Robeal, former Microsoft veterans with expertise in database applications, data synchronization, and mobile social networking. Raveable grew out of difficulties they’d each had in booking hotels quickly; they found they were sorting through dozens of reviews on multiple sites, without having a top-down view of how various hotels stack up against each other.
“We want to make it really easy to make a decision,” Vaughn says. “We were really frustrated by ‘Everything is 3.5 stars, everything is above average, everything is good.'”
The technology behind their approach is semantic analysis of text—an area that’s been in research for decades, but is increasingly being applied to Web search and corporate software. The goal is for the software to understand the meaning of sentences in user reviews—including the topic, the context, and the sentiment. So if reviews say the rooms are great, beds are comfortable, or parking is expensive, that’s pretty straightforward. But if they say the service could be faster, rooms get cold, the view is sick, or the place is in good need of repairs, say, the software relies on statistical models (trained and updated by the founders) to ascertain whether the sentiment is positive or negative.
Vaughn gives a sense of historical perspective, pointing out that Raveable fits into the trend … Next Page »