We know that this week is a United States national day holiday!, tens of millions of Americans went to the beach, walk down to the Lake, climb mountains, eating barbecue and drinking beer. For those holiday travelers, Airbnb is a Web site couldn't be better, because they can find the best waterfront home, where accommodation in the room, except you can relax and sleep, there are a variety of services, such as laundry facilities, hair dryer, wireless Wi-Fi, free parking, and more.
However, most people do not realize that when they landed at the same time on Airbnb and launched house searches, in fact, is very complex, and Airbnb, pressure will be great.
Of course, Airbn is not the search industry giants in the field, when we talk about search, Google and Amazon might every minute of "exterminate" Airbnb. But and these company different (worth a mention of is, recently years Facebook, and Instagram, and and Twitter are began put focus put to search above has), Airbnb is faced this a very unique of challenge, because it of search results does not just simple to looking for a website, a Zhang photos, or a paragraph products, Airbnb of user hope search to of results far more than these, like, some people hope in above looking for some right of housing, And compare this to redesign decorating their houses, some people just don't want in the middle of summer to relax and spend a lot of money in the hotel for two days, and some people simply do not want to be disturbed e-mail and want to enjoy the holiday weekend under their Lake House. However, all ... ... On Airbnb.
Yes, Airbnb has to do search work! they need to predict the housing within a database, to ensure that customers can find their ideal home. This means, Airbnb cannot simply confine all listings in a specific area, from this point of view, it is much harder than Google search. We know that in Google if you search for a Word, it simply displays all relevant pages are in random order, but not Airbnb, they can't take all the houses as random as a search result list.
"You always want to match supply and demand, and in our Realtor services under conditions provided by the Airbnb, the so-called supply is very unique. Because we search sources, real houses and House, "Airbnb Chief Technology Officer Mike Curtis said in an interview in New York," for the right guests, matching the right owner, it really can be said to be one of the most complex things! "
Machine learning
In order to solve this problem, Airbnb more and more start trying to use machine learning to understand the landlord and tenant's habits and preferences, in this case, of matching to each other will be more accurate. Each time, when a user searches the Airbnb accommodation information, the company will use a model to run a search, finding which landlords will most likely accept the user. In this model, there are many variables, such as guest residence, the latest potential book and homeowners the most recent time interval between long, and so on.
Through the test of this model, Airbnb's researchers found, based on the homeowner's custom search preferences, worked into the search results to improve 4% of actual booking rates. Therefore, Airbnb has decided to adopt this model. Decoding Tiger sniffing protection of intellectual
Meanwhile, Curtis said Airbnb also collected heads of other parameters, so that you can use on their machine learning model. For example, for the very limited amount of time to find housing for users know whether heads to accept last-minute booking can be very helpful, but can also provide similar services and companies (for example HotelTonight) competition. "We are beginning to transition to a Instant booking service, this is very important, because we understand the preferences of homeowners," Curtis said.
Search to fill the user
For Airbnb, the last piece of the puzzle is to use technology to understand user preferences, rather than simply filtering the user's choice. Actually, Airbnb can from many aspects to gets user preferences information, like according to they of click mode, system can learning a user special preference select which a room; also some user like will Airbnb and Concur with using, which is a specifically for Enterprise provides travel and costs spending management of system, this that is, through Concur access Airbnb of user, basically can judge out they belongs to business travel people, So they may need some exclusive facilities, such as indoor laundry facilities and wireless Wi-Fi network, and so on.
Of course, Curtis admits that the Airbnb forecast model in the user role of testing is still relatively limited, but he said that next year the company will continue to focus its work on searching this one, this is what they have to do the work. For today's technology companies, to more quickly achieve customer satisfaction, better the natural effect, such as Google on their search will choose the most favorable results at the top of the page; Amazon's recommendation engine also has its own, pushing some recommendation to consumers. More advanced search features, the more it will help the company gain customers, at least, can be faster than the competition, greater access to customers. Enhance the search function is not only on Airbnb, other online companies, are very important.
"For those of you who think in New York for the weekend, you can not only provides a place for users, or a place, so they will have no choice, you have to find a lot of places, and allow them to find the one you like best and the most appropriate House," Curtis said, "so we must become smarter!"
No comments:
Post a Comment