Ma Honeycomb to do International hotel reservation behind, is to become a "data

This article would like to respond to my former colleague Xiaomeng to do a new media experiment, the article is basically my talk with Chen Yu after the logic of his understanding and carding, did not carry out the wording of the arrangement. In order to make the article more effective, I would like to let you think about a question: why do I want to make international hotel reservation? What is his competitiveness? -Well, this is not to test everyone, just think first, the reception of information will be better.

Ma Honeycomb recently began to make international hotel reservations, the model is: MA Honeycomb is the decision-making entrance, through the UGC to help users make decisions, and then the user into the booking, Agoda and other booking sites, complete the transaction, divided.

Today, Chen Yu, the founder of Ma Honeycomb, talked about the logic of their hotel reservations, and here I share what I understand:

Ma Honeycomb do hotel reservations and OTA is not the same, OTA is the way to sell inventory, sellers perspective, the hotel put up to sell, and then produce a user a good quality or bad evaluation. But it is not the way to sell inventory, but the reverse from the hotel introduction, the expert perspective-to extract the previous traveler wrote in the introduction of the foot out of the experience and feelings, so that other users from the existing real experience to know a hotel, and then match these words into the book.

Why are the experiences and feelings left by the previous traveler important? Because these immersive experience to provide a sense of the later, such as the UGC can be extracted from the travel notes of a similar: "The hotel is very close to Shinjuku station, the Subway and JR line are very convenient, convenient to Shinjuku shopping." The hotel has underground access to the Shinjuku station, with a large box to walk will not be too hard "such a sentence, this is pure to do transactions ota difficult to do."

and to be able to extract from the existing long-winded strategy of the Effective hotel evaluation information, it is necessary to have a technical system of the honeycomb has a valuable data structure.

Chen Yu said the system was the "Raiders" they were doing at the start (the engine was fully upgraded in 2013) because they were the search engine-born team, and half of the team now has research and development technicians. In the search field, semantic analysis is a very important part-for example, "empty" and "more" these two relative words, in the financial field, and in the field of tourism, the meaning of the table is completely different. So they used semantic analysis in the strategy of data analysis, such as how to determine the user wrote to "Shangri-La" refers to the hotel or place? These require technology to extract and then structure.

I looked at the results of the MA honeycomb structured data, there are two more intuitive examples, one is the user can now directly in the introduction to see the hotel POI information, click on the link can be directly imported booking page. The other is the MA honeycomb for the restaurant reviews page, it extracts some of the most commonly mentioned words as a direct reference, such as "expensive Ah", "Taste General", "Ten years old", "" These words, the effect is close to the evaluation system Taobao.

So in this strategy under the engine, hotels, restaurants, attractions and so on the vertical data can be extracted from the previous user's strategy, made a separate product, used to do follow-up commercialization and other attempts.

So, what are the conditions needed to do a good job of data structure? Chen Yu said the first is to have a large number of UGC, the second is that the engine actually has a very high technical threshold. In chronological order, the leeches have gone through three stages, the first is the accumulation stage of UGC, do the community to start with the real content is very difficult, the second stage is to use the engine structure of data, structured well, the next stage is based on the data to do personalized recommendations.

"We used to be a social company, and we're going to be data company from 2014," he told his colleagues at the annual meeting. Data is not required to take a long time to fully integrate into the community to obtain, it is neutral and objective, like water and gas. With data, and then the user into a personality, there are characteristics, can be portrayed by people, you can provide them with personalized service.

Finally, I asked the next Chen Yu how to look at the travel from mobile end to share the community, what is the advantage of the MA honeycomb in comparison. Chen Yu's idea is that there is not enough UGC quantity to be able to enlarge the data. And the amount of accumulated UGC, in addition to the question of time, there is a problem that mobile is difficult to carry more services, he previously mentioned a point of view, this node PC and mobile competition is like aircraft carrier and fighter war, aircraft carrier is a very solid back-end, it can carry more data, More diverse service types and interactions that are not mobile-loaded, so a strong backend is missing from the pure mobile. In addition, he also raised a point of view, in the MA honeycomb, there are many users write travel notes, share travel experience, and then these experiences can guide others to travel, and constantly reciprocating to form a positive cycle, and other communities, almost no can form a positive circular capacity of the company.