Recently, Wasu has launched three blockbuster big data products of Aladdin Recommendation, data Compass and Magi Crystal Ball.
By collecting real-time business data from the support platform for Wasu digital TV services, The Viewership Compass uses real-time computing technology after a series of complex algorithms of big data platform to present the comprehensive condition of business operation in a high-efficient manner. It is featured with data authenticity, instantaneity and high-efficiency. The general condition of its web content operation can be known first hand, including real-time top lists of businesses such as the hottest films, TV series, children’s films and TVs, Video news, video column, live broadcast on its website and periodic top lists in prime time, days, weeks, and months. Viewership Compass can, for example, present with high-efficiency the comprehensive operating condition of all business platforms of the company within five minutes, allowing operating staff to know all this at once.
Aladdin Recommendation builds preferences of users by analyzing their behaviors such as browsing, video-on-demand and favorite, so it can launch dimensional services such as “What everyone is watching”, “ Guess you what you like”, “ Hot Top-N” and “ Associated content recommendation” that recommend films, TV series, video news and columns to users.
Magic Crystal Ball uses big data mining and analysis technology via big data real-time platform computing capacity to conduct comprehensive analysis on user behaviors on Wasu portal platform, make visual presentation, categorize and summarize user behaviors, and visually present user flow on the portal as well as users browsing through columns and pages, providing for operating staff, planner and layout designers all-dimensional business indexes for activities from real-time monitoring to periodic analysis.
Premier Li Keqiang has mentioned “big data” on several occasions since the hot word was first written into the government work report in March, 2014. The application of big data in the operation of new media becomes more and more important. Internet new media such as Youku, Tencent and Aiqiyi have seized the commanding heights on the technology front. They form the concept of big operation through basic functions of live broadcast, video on demand, Carousel, time-shifting and recording combined with interactive methods of user comments, scoring and voting.
Take Youku for example. The video recommendation of Youku involves several hundred parameters. Adjustment of the parameters demand manual adjustment of more than ten or tens of parameters. What’s more, billions of data is involved in the data model of everyday video recommendation, making a small parameter adjustment likely to bring an increase of several million video viewers. The American drama House of Cards stands as the first strategic application of big data analysis. All the episodes are released on the website by Netflix for the view of subscribers, totally overturning traditional episode release model of one episode per week. The result shows with three hit factors of Kevin Spacey, David Fincher and BBC production rolled into one, considerable audience can be attracted. It can be seen the viewing behavior pattern has changed in which increasingly people no longer expect to watch the latest episode in front of the TV at a fixed time of a fixed night 30 years ago, instead, they select a convenient time to watch all the episodes on a convenient device.
Generally speaking, for poor user viscosity and low user loyalty in the industry of new media, every media, besides elbows for upstream content resources, employs big data to collect, compute and analyze at any cost, providing effective support for all businesses in big operation. The development of big data usher Internet media operation into an advanced level. It is big data that creates the advanced level of media operation.
Wasu Media has accomplished big data infrastructure from scratch on the strength of TDH Hadoop big data platform developed by Transwarp. It has achieved data docking with portal,CLPS , and basic applications of user list and user recommendation. It can be regarded that Wasu Media has been the leader in big data field of radio and TV industry.
However, as the difference between technology and network in radio and TV industry has been abridged, user experience stands as the most important factor in users choosing service providers. But the strong blow from Internet new media makes it imperative for radio and TV industry to narrow technology gap. As the leader in this industry, Wasu must remain fully competitive in user experience. So it must dig deep into user values, analyze user preferences, make its current big data platform stronger and bigger, maintain technical advance, and remain competitive in the market.
In 2014, Wasu Media big data platform has successfully dealt with the four challenges of stronger service performance, reduced time of data mining and analysis, scale-out of big data business supporting functions and service quality data analysis. After two years, as changes take place with each passing day in big data field, Wasu will face new market, technology and challenge. So there comes the three-phase planning of Wasu.
It can be learned from analysis of Wasu’s condition, Wasu’s log collection has basically accessed to interactive TV domain service data, Internet TV and partial maintenance data (call center and broadband center). But granularity of log collection is too thick, making it unable to provide support for finer analysis. So the first task in the planning is to strengthen data collection and analysis.
Current log only includes the basic log on users watching VODs and live broadcast, which are unable to reflect the real preferences of users and needs to be more precise and accurate. Finer log includes clicking distribution when users browse pages, user page stops market, exit points, operations of fast-forward, fast-backward and exit when users are playing videos.
In respect of data collection and analysis, the behaviors initiated by users tend to be more valuable than information passively collected. The log on behaviors initiated by users includes comments actively made by users, interactive behaviors such as scoring, breakdown information reported by users through call center, content review from network users and scoring data fusion.
Besides, the building of platformization of data collection is also urgent. The collection of massive data, cloud-host and service have become a hot business, such as AMAZON’s REDSHIFT. Wasu can consider platformizing its current collection capacity based on its big data platform to provide data collection and services.
Meanwhile, the transformation of functional system to operating platform is also necessary. Among the three versions of Wasu big data platform, V1.0 achieves the infrastructure building of Wasu big data from scratch, and accomplishes basic recommendation, index, top list business and access to partial business system. V2.0 strengthens the previous functions of basic recommendation, index and top list business, achieving tools that presents analytical results of Crystal Ball and Viewership Compass, as well as docking with part of the operation and maintenance data and streaming real-time computing analysis.
There are problems in four respects of current platform needing to be resolved and optimized.
First, degree of integration is not enough. Currently only functionalization is achieved. It simply heaps all the functions as a system according to needs, lacking in centralized platform of high controllability and loosely coupled functional components.
Second, it lacks comprehensive analysis. In the initial phase data of all businesses serves the businesses discretely, causing insufficiency in completeness and authenticity of analysis as well as in comprehensive analysis on total business data.
Besides, it is in absence of openness. Whether for market demand or the need of policy development, openness and sharing stand as important features of platform. At present, big data systems are unable to achieve openness in capability so that external data cannot be imported while internal data cannot be shared out.
At last, it has a high threshold in extensibility. The horizontal business extensibility cannot be achieved at a little cost for insufficiency in functions of components. Thus Wasu’s big business operation ecosystem cannot be supported by it.
Therefore, V3.0 should transform to an operating platform on the basis of its original functionalized system. It should achieve such functionalized component architectures as computing center, collection and service, horizontal extensibility including business role decentralization on management end and partition on the data end, standardized business data models and data interface for external services, as well as unified big data analysis.
The last challenge it faces is the need of richer business types. Current big data system has achieved basic services including recommendation, index and top list and the operation and maintenance of call center and broadband faults, increasing conversion rate to some extent and saving human cost in this regard. But it lacks intuitive operational instruction and businesses that can bring about user value transformation, with insufficient penetration into businesses that can reduce operation and maintenance costs.
Efforts to expand businesses include bringing intuitive data guidance for the arrangement of operation through big data comprehensive analysis. The guidance covers hit volumes, popularity of network-related content, popularity of region-based content, popularity of users’ page positions, browse tracks on users’ pages and statistics on users’ viewing behaviors. An effective guidance on periodical marketing is formed by means of statistics on marketing subjects and activities in different periods. The statistical analysis includes content popularity, preference for pages and others.
Expansion of businesses that can transform user value includes tracking users’ overall behaviors from the time they power TVs on to the time they power off TVs, detailed comprehensive analysis on users’ operations in every link, and depicting more complete user images. It carries out further excavation on the basis of current content tags system, and combines internet new media tags to convert content tags to real operations. With the combination of complete user images and content tags, it launches new businesses such as customized ads, recommendations and subscription promotion to transform user values.
The business expansion of cost reduction in operation and maintenance includes enriching current operation and maintenance data such as administrator operation analysis, server fault-reporting analysis, and database fault-reporting. All the data above is gather in big data center which can, after comprehensive statistical analysis, provide global view of classified operation and maintenance. It can help locate types of faults that frequently occur, and fault-prone areas, solving problems from the root and reducing repetitive maintenance cost.
Generally speaking, the objective set out for V.3 big data platform is to expand the spectrum of big data collection, achieve operationalized platform and enrich business types.
For many businesses including radio and TV operators, command of refined user needs is an irreversible trend for future development. In this respect, the advantage of Hadoop big data is incomparable. The big data platform created by joint efforts from Wasu Media, Transwarp and Sihua Tech has endured the test of time in the past two years, during which Wasu has always taken the lead in this industry. Now the plan to build V.3 big data platform again proves that the tide of the big data age still develops in full swing.