Shulou(Shulou.com)06/02 Report--

In the 21 century, new technologies of the Internet emerge one after another. With the development of big data, cloud technology and the whole computing power, the research and application of artificial intelligence technology is also growing rapidly, and has made outstanding achievements in voice, image and natural language. More importantly, the government is also making great efforts to develop the digital economy, including the addition of "data" as a factor of production for the first time at the fourth Plenary session of the 19th CPC Central Committee, which further reflects that with the acceleration of the digital transformation of economic activities, the multiplier role of data in improving production efficiency is highlighted, and it has become an important change in new factors of production with the most characteristics of the times. GE Tui, as a professional data intelligent service provider, was established in 2010. In the process of the development of big data's technology, it has gained some achievements and gained its own experience.

In March 2019, GE Tweet (Daily Interactive, 300766.SZ) was listed on the gem, focusing on data intelligence. The company has actively laid out in this field, and has explored and summarized the "three-step" data intelligence methodology of data precipitation, data mining and data application in practice. With regard to the bottom layer, a push provides four kinds of developer services: message push, user profile, application statistics and one-click authentication, and precipitates a large amount of data at the same time; the middle layer is a push engine layer, through desensitization, screening, cleaning and sorting of the data, and after in-depth mining modeling, independently build a "push big data platform". The top layer is the data application layer, which provides a variety of big data services, such as brand marketing, risk control services, population spatial planning, public services and so on.

This paper will be combined with the practice of individual promotion, mainly around three aspects: the characteristics of the era of data intelligence, the practical challenges faced by enterprises, and the idea of building data intelligence system.

Characteristics of the era of data Intelligence

The era of data intelligence has come. As described in the big data era, we have found and confirmed that big data has a great influence on our thinking, business, management and so on.

Thinking change

In terms of thinking change, we think that the characteristics of the era of data intelligence can be summarized into three points: more, more complicated, and more relevant. More means that people no longer rely on a small part of the data, but use all the data and leap from the third paradigm to the fourth paradigm.

So what is the difference between the third paradigm and the fourth paradigm? Take "the reason for the formation of haze" as an example. The haze research process of the third paradigm is as follows: first, ask questions. For example, want to know what haze is? How to prevent it? Secondly, put forward the theory. According to the existing mechanism, the formation of haze weather is not only related to the source and atmospheric chemical composition, but also to meteorological factors, including topography, wind direction, temperature, humidity and so on. The number of parameters is beyond the scope of our conventional monitoring capacity.

To this end, we can only remove some parameters that do not seem to be important, retain some simple parameters, and put forward a theory. Then collect the data, simulate the calculation, and modify the theory according to the calculation results. Finally, the results with high credibility are obtained to predict the weather of haze.

Correspondingly, what is the research method of the fourth paradigm? Its first link and the last link are the same as the third paradigm, but the order of the middle two links is opposite, that is, in the fourth paradigm, we have to collect data before forming the theory.

After the question is raised in the first step, the first thing we do is not to create a theoretical model, but to collect all the data that may be useful, and then on this basis, through the method of machine learning, or the method of artificial intelligence, to find out the factors that have great influence on the formation of haze, and then put forward the relevant theory. Finally, the prediction is made and the results are verified. Thanks to the emergence of big data, the third paradigm can leap forward to the fourth paradigm, which has also brought about changes in the whole way of thinking and methodology.

The second feature of thinking change is more complex, that is, the shift from accuracy to probability. Accuracy, as the product of the era of lack of information and simulation era, requires data to ensure the quality and reduce the error. In the era of big data, the massive data made it impossible for us to verify the accuracy of the data one by one. More often, we find out the probability behind the data through the distribution of the whole data, and then find the useful data and eliminate the useless data.

The third feature of thinking change is that it is more relevant. The relationship between data is not causality, but correlation, and the core is the prediction based on correlation analysis.

To sum up, the characteristics of the change of thinking in big data's era can be summarized as follows:

Change the mode of operation and use all the data collected instead of samples; do not focus on accuracy; accept confusion and errors; and focus on analyzing relationships rather than the reasons behind predictions.

Business change

In terms of business change, there are three main characteristics of the era of data intelligence: everything can be quantified, unlimited possibility of innovation, and the choice value of data.

"everything is quantifiable" means that with the development of the era of data intelligence, we will find that everything around us is generating data, that is to say, the real world we live in has a corresponding relationship with the information world. In the future, everything in our physical world will correspond to one by one in the digital twin world.

The true value of data is like an iceberg, the primary value is only the visible part above, and there is "unlimited innovation possibility" behind it. After completing the direct business use, the data seems to be useless, but once combined with other industry data, we will find that its synergy is very strong and can create great use value. In other words, business data that seems worthless at present may play an important role in the future. Therefore, we suggest that enterprises or companies with rich data, it is best to save business data in some way from now on, such as using the data lake scheme.

The value of data is the sum of all its possible uses, in the face of these infinite potential uses is like a choice, the sum of these choices is the value of the data, that is, the choice value of the data.

To sum up, in terms of business change, the characteristics of the era of data intelligence can be summarized as follows:

The choice value of data means infinite possibilities.

2. The age of mathematical intelligence requires us to treat data differently from traditional assets.

3. the innovation of data means a lot of uncertainty.

The practical challenges we face

The essence of the challenges we face in the mathematical era lies in the conflict between the requirements of data organization and management (focusing on stability) and the innovative needs of the business (focusing on flexibility).

Several problems to be solved in using data Core

Data is unknowable: users do not know what kind of data big data platform has, nor do they know the relationship between these data and business. Although users are aware of the importance of big data, it is not clear whether there is key data in the platform that can solve business problems and how to find it.

Uncontrollable data: uncontrollable data has been a problem since the traditional data platform, especially in the era of big data. The lack of a unified data standard makes it difficult to integrate data, and the lack of quality control makes it difficult to use a lot of data because of its low quality.

Data is not desirable: even if users know what kind of data their business needs, they can't get it easily by themselves. In fact, data acquisition requires a long development process, and the long demand response runs counter to the goal of rapid issue of problem solutions in big data's era.

Data can not be connected: in the era of big data, although enterprises have huge amounts of data, the relationship between enterprise data knowledge is still relatively weak, and the data and knowledge system have not yet been associated. In addition, it is difficult for enterprise employees to achieve the rapid conversion between data and knowledge, unable to carry out in-depth exploration and mining of data, which makes the deep value of data difficult to highlight.

We collect data problems within the company and find that there are several difficulties: slow business response, frequent data quality problems, difficult data use and slow data fetch, low development efficiency, high trial and error cost and repeated construction of data capacity.

Thoughts on the Construction of data Intelligent Technology system

Overall goal

1. Agilely support the innovative needs of business units and build service capabilities that respond quickly to business needs

2. Connect the data in different fields in real time to reflect the maximum value of the data.

3. Manage data as assets.

In most cases, we promote the construction of the company's data intelligence system through the needs of the business, and its direct value is reflected in cost saving, efficiency improvement and quality improvement.

Ideas and principles of construction

1. Mainly for internal customers, especially the company's R & D personnel and modeling personnel, with the goal of improving the efficiency of business development

2. Do a good job in metadata and consanguinity management, improve the degree of data governance, so as to ensure the quality and security of data.

3. Priority should be given to the construction of high-reuse capabilities such as public service capabilities, such as data extraction and analysis speed, data governance platform and data development platform.

4. In principle, the data capability is built by a team that is familiar with the business in the corresponding field and has certain technical accumulation.

5. Capacity-building needs to focus on several major standards: stability, easy operation and maintenance, operable and auditable.

In terms of capacity-building, the company can set up a three-tier structure: the bottom layer is the technology center; the middle layer is the data center; and the upper layer is the business system. It should be noted that having a platform does not mean that the problem is solved. We think the most ideal way is to combine platform with human ability. Platform precipitation proves the ability of reusability, while people respond more to the needs of innovation, using knowledge creation tools and improving the platform. This is also a spiraling process. The platform needs specialized people to operate and promote; the business needs people who can use the platform and can quickly generate solutions to ensure good communication and cooperation with platform personnel.

Based on this idea, a push built such a system on the company's organizational security: the upper layer is currently a virtual data center department, which will become a physical department at the right time. The architecture group and the technical group participate in the construction of the data center. In addition, we send some technical personnel related to data to the business department, so that we can not only better apply the data to the business, but also let them feedback the results and problems of the business unit, so as to form a closed loop, which we call DO (Data Owner).

The development of the Internet has brought everyone into the era of big data, and the era of data intelligence is an important stage of development in the era of big data, with both opportunities and challenges. GE push will actively seize opportunities, meet challenges, and constantly explore the combination of data intelligence and industry applications. Innovative technologies will provide developers with enhanced services and customized big data solutions for various vertical fields such as mobile Internet and brand marketing. In the future, a push will continue to use the power of data, join hands with more industry partners to create a data intelligence win-win ecology!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.