Video site "Encounter" Big data

Intermediary transaction SEO diagnosis Taobao guest Cloud host technology Hall

Reading: Big data as a hot word nowadays, is not as simple as people see. When the big data encounters the video website, its utilization direction?

With Netflix (an American online movie rental provider) the first original series "Card House" hot, the use of large data to make a video site operation is a hot topic. Behind the show, "Big Data" almost runs through all the parts of the card house, including production, marketing and distribution.

In the first quarter of the "card House", Netflix's earnings showed 3 million new users, and Netflix's share price soared 26%, with the market capitalisation doubling over last year's lowest value, leaving the company, which had already been at the bottom of the valley, to roll over.

Netflix in the industry in the heat wave of the domestic video site has also had an impact. Archie, chief executive of Arts and crafts, Gong recently said that this year's love of Qi art will be through large data analysis, a commercial micro-film into a big film. Other domestic video sites have also turned their eyes to big data, hoping to make use of big data nuggets.

Customized based on user preferences

Big data is not as simple as people see when it is a hot word. When the big data encounters the video website, its utilization direction?

"Video site can be mainly from three aspects of the use of large data, the first is the basis for procurement, the second is the basis for user operations; The third is to support original self-control, behind the data engine. "Youku's President Wei that.

With the success of the card house, it has become a new way to make video websites profitable by making homemade dramas with the advantages of low production cost and the only channel to play. "Analysys International analyst Zhang Yi told the China News and publication newspaper," The domestic main video site has accumulated a large number of user data, through the analysis of user data can accurately locate the site user's behavior preferences, so you can also shoot homemade drama. "User data analysis is actually a human analysis, if a user watching video to see a favorite star to take a move mouse or pause behavior, then this node behavior is meaningful." "Zhang Yi that, like NEXFILX through the analysis of user behavior data to determine the" card House "in line with the needs of users, large data analysis for video site-made drama has a certain guiding role.

Archie also said that the large data related to the video site is mainly user media search, watch, share video and other production of the log, but also contains users in Baidu generated search behavior. Archie in the user's film and television content recommendations, precision advertising and film shooting will use large data.

Personalized push of Thousand Faces

In many cases, users log on to the video site do not know what they want to see, they often input content in the search box can not find the video, but simply rely on the search will still give users a sense of confusion. How to reduce the user face mass video content more and more high choice cost, provide more active personalized service, become the challenge of video site at home and abroad, in this regard domestic video site has also tried.

Youku has a personal User Center, registration and login system will be based on user browsing and viewing behavior recommended 20 videos, if you feel bad can also change. In addition, Youku tries to achieve a multiple-screen cloud recording that allows users to watch the same video continuously on different terminals.

A few days ago, Archie also made a comprehensive revision of the PC client, trying to rely on large data analysis to achieve "thousand people thousand Faces" home page personalized content recommended. After the revision of Love Art video, home focus, hit the strong file, entertainment gossip, Animation park, high-definition movies and other navigation content as well, but each user in different regions, different times to get the recommended content is no longer the same. Tangxing, chief Technical officer of Archie Arts, said: "This is more than 35% percentage points for user-recommended content, and the recommended amount of playback is more than 50% of total traffic." ”

Personalized service is important to improve a user's viscosity of the site, and through the user browsing, viewing the data left to achieve personalized custom service is theoretically a good idea, but Zhang Yi also believe that the current data mining technology has not reached such a precise level, some of the recommendations of the site are based on structured data of the primary application, "If a user takes action while watching video, this behavior can be recorded, but in the analysis, the semantic understanding of user behavior and accurate analysis is still difficult." At the same time, large data mining needs to invest in higher human and equipment, technical costs, the current domestic video site attention and investment level is still not high, large data easily become a marketing gimmick.

Accurate personal ad Placement

In the second quarter of 2013 China online video Market scale quarterly monitoring report data show that the Chinese online video advertising market size of 2.14 billion yuan, an increase of 49%, the chain growth of 39%. The main source of revenue for the online video industry remains advertising, which accounts for up to 75.2%. The rapid growth of the online video advertising market is a reflection of the increasing value of the online video business, says Eric. The growth of the future online video advertising market will be driven by rising advertising prices. Thus, advertising for video sites is the main way to profit, through accurate advertising to attract advertisers is the video site to achieve the goal.

Zhang Yi introduced that the user's behavior data analysis can be known to some of the user needs characteristics, thereby improving the accuracy of advertising. "If a user stops at a certain screen while watching a video, it is likely that there is a star or object in the screen that attracts him or her, and the accuracy of the advertisement will be greatly enhanced if the advertisement is inserted with the star or related items." "he said.

The current video site on the basis of classification of people based on content classification, user browsing records and historical click behavior and other factors. Archie Art Aspect thinks, this kind of user "portrait" The advertisement form may in the brand advertisement display to play the very good launch effect. "Such as Nestle coffee a network video patch advertising, Han is the leading role in advertising." Han's loyal ' fans ', embedded interactive ads can highlight Han one, users click on small patches into, click Interactive Ads, users can play a small game, or see more information on advertising products, and by inputting text comments, in micro-blog interaction. "Tang Xing said, fully understand the user can bring the effectiveness of advertising."

But at the same time, Zhang Yi also told reporters that through large data analysis of user behavior to achieve advertising still have a semantic understanding of the problem. In addition, the privacy protection of user information, the open cooperation of data between different websites is the bottleneck of large data utilization. "Big data is a very complex thing, even though everyone knows it's promising, but the real use is not yet well formed in the market." He said.