Recently, iResearch released the 2017 "China Business Intelligence Industry Research Report", focusing on the industry application of artificial intelligence and business intelligence, that is, how artificial intelligence technology can be applied to business intelligence decision-making, and how to realize the intelligence and automation of business operations. .
The report pointed out that the demand for refined operations of Chinese enterprises is exploding, and the requirements for business intelligence solutions have increased, especially in the fields of finance, e-commerce, logistics and travel. Shift in decision-making. Technically, the future of business intelligence will focus on a single technology to the integration of various disciplines, branches, and algorithms. In experience, the understanding of scenarios by enterprises and technology suppliers is the key to industrial upgrading. The implementation of business intelligence is a systematic project, and the engineering practice capabilities of enterprises need to be enhanced.
1. Overview of the development of China's business intelligence industry
Benefiting from the national economic development and policy support, in the business intelligence industry, China and the United States are already in the first phalanx, and their development trends are different. For example, Chinese enterprises prefer an integrated solution from the underlying database to the front-end application. Sailor Brother agrees. From the west to the east in China, the level of informatization varies in different regions. However, in recent years, everyone has attached great importance to the experience in the industry and pursued integrated solutions. The core reason is to seek data. To optimize the solution and get a whole set of systems is equivalent to hiring a professional consulting company to sort out the business. In terms of technology, AI and big data are booming, coupled with favorable policies, China's business intelligence is showing an upward trend as a whole. However, due to the moderate economic growth rate, the dividend period of rough and rough operation has passed, and the demand for refined operation is exploding.
Meanwhile, iResearch briefly described the business intelligence industry chain in this report. The application of intelligent technology in business scenarios, that is, the midstream and downstream of the industrial chain. Regarding the upstream of the industry chain, traditional IT vendors and cloud service vendors can provide computing, storage and other infrastructure support for technology, product and service providers. Information systems such as ERP and CRM can help companies effectively record their resources and business data, and integrate data. The third-party data of the user can enrich the data dimension of intelligent analysis.
FanRuan
is a big data BI and analysis platform supplier and is in the middle of the industry chain. It can provide both standardized BI products and industry solutions.
2. Analysis of the future core technologies of business intelligence
The report believes that the core technologies of future business intelligence are machine learning, knowledge graph, and operations research. The so-called machine learning is to automate itself. Since 2015, neural networks have risen again in the name of deep learning, greatly improving the accuracy of perceptual intelligence. The core machine learning algorithms include classic algorithms such as support vector machines, naive Bayes, decision trees, and neural networks, and include hot technologies such as deep learning, reinforcement learning, and migration learning.
The knowledge graph was officially proposed by Google on May 17, 2012, and its predecessor can be traced back to the Frame Network (Semantic Network) in the 1960s. In terms of coverage, knowledge graphs can be divided into general knowledge graphs that are relatively widely used and industry knowledge graphs that are dedicated to a specific field. The general knowledge graph focuses on breadth, emphasizing the integration of more entities, and is mainly used in fields such as intelligent search. The industry knowledge graph needs to take into account different business scenarios and users, and usually needs to rely on the data of specific industries (such as finance, public security, medical care, e-commerce, etc.) to build, and the attributes and data patterns of entities are often rich.
Operations research is the science and art of using quantitative analysis to make decisions and optimize. It provides wisdom for management decision-making and solves management decision-making problems with its own wisdom. The era of big data gives enterprises more massive, more dimensional, and more timely full-sample data, and also brings new business, new scenarios, and new constraints for industrial practice. These new problems bring fresh nourishment to the classic theory of operations research , which promotes the continuous emergence of new models and methods of operations research. Amazon in the United States will have an operations research team of hundreds of people responsible for revenue management issues such as logistics, warehousing supply chain optimization, and commodity pricing; Google's search engine is developed by people with computer science backgrounds, and now Google also has a dedicated operations research team. Solving various problems such as corresponding advertisement clicks and street view path optimization, all rely on operational research optimization to do fine operations in the era of big data. The explosive growth of the domestic mobile Internet has brought a large amount of data accumulation and precipitation, which to a large extent supplements the original PC-centered IT information system. The data recorded by users in the process of using mobile services has become an intelligent decision-making system. The basis for improving operational efficiency, e-commerce revenue management, supply chain optimization, online car-hailing route planning, dynamic pricing, financial risk management, and organizations in various fields are full of imagination due to the huge domestic user market.
However, from the perspective of product technology development, Gartner has different views. They believe that in addition to deep learning, future business intelligence technologies include data preparation automation and natural language processing.
Data science tasks such as data integration and data preparation will be increasingly automated, which will greatly improve the efficiency of data scientists. Automating manual tasks such as data integration can help improve the efficiency of professional and civilian data scientists and alleviate staff shortages. Gartner predicts that by 2020, more than 40% of data science tasks will be automated, leading to increased productivity and wider use of citizen data scientists.
Natural language understanding is an important direction in the field of computer science and artificial intelligence. It studies how to make computers understand and generate the language that people use every day, so that computers can understand the meaning of natural language. BI's natural language processing technology will be able to express the results of data analysis to users in a narrative way, which is easy for users to understand. Users will analyze and query through search and natural language, like siri-like interaction, without having to query through experts or computer language. But Sailor Brother doesn't agree with personal voice assistants. The Chinese are not used to it yet, so he always asks about anything by voice. Asking "Wu Fan, tell me the sales of FineBI in Shanghai in the past two years" in the office will probably be regarded as a lunatic.
3. Typical applications of business intelligence
1. Advertising and marketing: Precision marketing is responsible for attracting customers, and personalized recommendation promotes survival and retention
2. E-commerce: The essence of revenue management is optimization, and intelligent revenue management helps companies increase revenue without increasing traffic input
3. Transportation: Minimize distance and travel time through artificial intelligence + operations research
4. Supply chain: Improve the efficiency and flexibility of the supply chain system through big data and optimization technology
5. Financial risk control: use data and technology to improve the accuracy of risk control, and deploy the whole process of risk control
6. Intelligent customer service: upgrade from manpower-intensive to man-machine hybrid mode
4. Challenges and future of business intelligence
Since artificial intelligence has become the focus of industry, academic circles, investors and the media, the public has paid particular attention to technologies such as deep learning. However, in industrial practice, the understanding of specific business scenarios and the definition of practical problems are equally important as the model and algorithm to use. The former largely determines whether the latter can effectively reduce the operating cost of the enterprise or help related businesses. Increasing income is the key to enabling technology to be implemented and the industry to be upgraded. In AAAI2017, Gary Marcus, director of Uber's artificial intelligence laboratory, said that the current rapid development of deep learning and other technologies may only be approximating a local optimum of general artificial intelligence, and such an approximation may make us miss those who are really better. Methods to achieve general artificial intelligence. Therefore, before using technology to solve a problem, one should never preconceived that a specific machine learning algorithm should be used, but should first analyze the business scenario and grasp the core problem elements, which is to make the optimal technology choice the premise.
The implementation of business intelligence business applications needs to be based on perfect data integration and management, and then the corresponding algorithms and models will convert the data into visual business rules based on an efficient computing framework, and further drive or directly generate enterprise decisions. Therefore, business intelligence It is a systematic project. Algorithm design, architecture construction, system coordination, process control, quality supervision, crisis management, etc. are indispensable. Project engineering experience is very important. On the other hand, analogous to Salesforce, the world's top SaaS company, the general functions of its products only account for about 50%. There are still a large number of suppliers and their own service teams behind the products to provide customized services in combination with customer differentiated scenarios, so it is still in the early stage of business intelligence. , For a long period of time, the service mode will still be dominated by customized solutions (especially for large enterprises), supplemented by standardized products such as SaaS, and in some scenarios, PaaS services will be used to access customers ERP, CRM and other information systems can quickly and cost-effectively empower enterprises with business intelligence.
This article first published CSDN:
http://blog.csdn.net/yuanziok/article/details/73613324
For more big data information and enterprise cases, please follow: Official Account-FanRuan Data Application Research Institute, ID: fr_research