Thoughts on Business Intelligence Philosophy

Http://qkzz.net/magazine/1005-2348/2007/07/851623.htm

To some extent, the current development of Bi is subject to the unsatisfactory implementation results caused by the deviation of users' and developers' understanding of Bi. The following lists common problems of actual Bi projects. These problems cannot be solved in the current Bi system.
For users, the question is: What is the essence of Bi and how should we understand bi? What kind of bi products should we choose to meet the needs of our enterprise in the face of a large number of bi products and different functional versions? What Should bi construction be like? How to assess the difficulty and depth of the enterprise's Bi and consider funding and cycle issues accordingly? What should enterprises do in the future for Bi?
Developers should think: how can we integrate our concepts into user awareness? How can I introduce Bi to each other to ensure mutual harmony? How can we work with users to explore their needs and smoothly recommend corresponding products and functions? How can customers understand the difficulty of this bi project without making a "bargain" on expenses "?
These problems are widely stored in various Bi projects, mainly because of incomplete BI systems and propaganda and guidance from bi vendors and agents.
　　
(1) incomplete and one-sided Bi system leading to cognitive bias
　　
Compared with the definition of Bi, there are few discussions on the Bi system at present. The following describes the BI systems of some famous international manufacturers. It can be seen that the current Bi system focuses on technology in terms of cognition, and "blindly touching the image" will inevitably lead to some direction errors.
　　
Microsoft Bi SYSTEM
Figure 1 is Microsoft's Bi system. It consists of three layers: database, and application function. A total of seven applications are listed, such as Bi joint operations, visualization, geospatial analysis, click flow analysis, retail and marketing analysis, project management, and data analysis. It can be seen that the core meaning of this system is to implement various applications based on the data in the database, reflecting the general idea of Bi.
　　
Figure 1 Microsoft's Business Intelligence System
　　
IBM Bi SYSTEM
Figure 2 is an IBM Bi system with underlying components: software, hardware platform, middleware, security, and metadata management. It reflects the workflow and data stream in Bi, from right to left is the data source, data integration, analysis and interface. The IBM Bi system details the data transformation and application process, as well as the data transformation and application assurance process. The core is to solve the data usage and data accuracy issues, which are obviously technical thinking.
　　
Figure 2 IBM Business Intelligence System
　　
Bi system of the us data warehouse Research Institute
Figure 3 shows the Bi system proposed by the Data Warehouse Research Institute. The Institute compared Bi to a data refinery, believing that Bi is produced by a transaction operation-type system and transformed by a data warehouse and a series of tools and models, implement data to information, knowledge, policies, and then to action, and guide the cyclical process of the transaction system. This system focuses on business needs and converts Bi results into actions in the OLTP system to achieve a loop. However, it is still based on solving the single problem from data utilization to data utilization, the shadows of technical thinking are still very heavy.
　　
Figure 3 business intelligence system of data warehouse Research Institute
　　
Bi system of domestic researchers
Yang Lin, a Chinese researcher, believes that the Bi system framework consists of four layers, as shown in figure 4. The four layers form a pyramid. The bottom layer is the information system layer, the data analysis layer, the Knowledge Discovery layer, and the top layer is the strategic layer. The framework focuses on the management layer, but its applicability is insufficient.
　　
Figure 4 bi framework
　　
The above system focuses on the technical point of view to reflect the Bi process from data to application in a slightly different degree. IBM has added control and security for this process. The US data warehouse Research Institute shows the association and impact of Bi results on policies and actions. These BI systems are not comprehensive and close to application, or users cannot understand them in depth. They simply use bi as the process of using data; or developers can only know the implementation process of Bi, but cannot understand the implementation results and causes, so that they cannot communicate smoothly between users and developers.
　　
(2) Promotion by Bi manufacturers and their agents focuses on or induces greater cognitive bias
　　
Bi products are divided into data warehouses, OLAP report tools, ETL tools, and data mining tools by function.
Currently, no products have been implemented throughout the Bi process. Each vendor has only one or more products in the process. For example, SPSS products focus on analysis and mining. Bo products focus on OLAP analysis and report Presentation (Business Objects; Crystal Reports), and Cognos has analysis and presentation tools. On the other hand, even products in the same Bi process are different from each other. Therefore, in the face of the word "bi", users who do not know Bi are inevitably lost in the promotion of various manufacturers.
It can be seen that due to the "Ignorance" of Bi users, the differentiated strategies of different Bi manufacturers form "differentiation based on differentiation" rather than "differentiation based on standardization ". As a result, communication between users and developers is blocked, or even unilaterally induce user communication, which is obviously not conducive to the development of the Bi industry.
Therefore, building a system with philosophical thinking is the fundamental way to solve the current Bi system problems.
At present, the Bi system only focuses on how to do Bi technically. Obviously, this method only improves a hole on the wall, no matter how well it is perfected, it is difficult for users and developers to unify their full understanding of each other and to jointly promote industrial development. In this situation, we need to draw on the "existing holes" and then create another hole in the wall and connect it with the "existing holes, form a New Bi system that facilitates communication between users and developers, and interpret, describe, and operate to achieve a common understanding and understanding. This not only makes it easy for users to identify the focus and guidance of various manufacturers, but also allows developers to better understand the correct user cognitive structure of Bi so as to facilitate communication and transactions, ultimately, "differentiation above standardization" is achieved to reduce communication costs.
The interpretation of the system in CI Hai is a whole composed of several related things that are mutually restricted. According to this definition, the Bi system is the whole of Bi-related things that are interconnected and mutually restricted.
How to build a system in which Bi involves intercommunication and mutual control of related things? The author believes that only by interpreting and constructing a unified system of ontology and methodology from the perspective of ontology and methodology can we clearly describe bi-related things and their relationships. At the same time, the regression of philosophical thinking, this allows users and developers with different experiences, different levels of knowledge, and different levels of practice to share and explore Bi based on the same starting point of understanding, so as to eliminate Bi misunderstanding and reduce Bi transaction costs, promote the development of the Bi industry.
Philosophy is not only a theoretical and systematic world view, but also a methodology for observing, analyzing, and handling problems. The so-called world view refers to a person's fundamental view of the whole world. It also refers to the theory of existence, its nature, and its laws. It is a high abstraction of existence, called Ontology. The ontology is called ontopology, the basic meaning is "Yes" and "exist ". Ontology solves the question of what is and what is the sum. It is the material object of methodology and epistemology.
Methodology refers to the existence of traits, attributes, functions, functions, forms, and methods of material objects, for example, descriptions, explanations, and explanations of the absolute dual nature and unity of opposites, movement, structure, function, form, change, transformation, development, universal connection, quality mutual change, and negation of negation, it is the theory or theoretical system that people use to understand the world and transform the world. It is also the basic principle and philosophical basis for people to observe, analyze, and handle various problems.
Methodology solves the problem of how things are moving and the results of motion. Philosophy is the unity of ontology and methodology. In general, what kind of ontology is there as well as what kind of methodology. Ontology is the theoretical basis and philosophical basis of methodology, and methodology is the concrete application and embodiment of world outlook.
Therefore, to build a Bi system starting from philosophy, we need to unify the ontology and methodology. Starting from the subject, the Bi system should solve the problem of understanding Bi by users and developers (this article mainly refers to secondary developers and integrators, that is, we need to clarify what Bi is (including why we need to know Bi in this way), how Bi is used, how Bi is implemented, and how it is interconnected and mutually constrained.
In addition, it is necessary to provide a common platform for users and developers engaged in Bi research and practices. The Bi system in Figure 1 is to form a "body, use" relationship. The "body" fully embodies the ontology, which is suitable for users and developers to jointly understand and communicate with each other. The "use" includes what to do and how to do it. The former is suitable for users and developers to jointly explore the project results, the latter is mainly suitable for developers. Of course, users are familiar with the results of Bi projects and understand the methods and processes of developing bi, which helps them understand the difficulty and duration of the project and facilitate budget.
Compared with the Bi system built by philosophical thinking and the current Bi system, we can see that the current Bi system only solves one problem, that is, how to do Bi, this is only part of the "use" in the body usage relationship. It can be seen that the current Bi system is incomplete and one-sided, so that it is difficult to solve what the user expected Bi is and why, what kind of Bi is, what kind of Bi should be made? How to Improve Bi in the future, so it is impossible to smoothly communicate with developers and users.
　　
　　
Bi's "body" Framework
　　
Based on the above-mentioned relationship, the author has established a "body" framework system for Bi. The "body" Framework refers to the Bi system, which consists of three parts: "three-dimensional architecture", "three operating modes", and "four-step development theory. 1. As shown in, Bi's three-dimensional architecture is the foundation and "body". Bi's three operating modes and four-step development theory are built on the basis and are "use "; these two parts constitute the "body, use" Relationship of the Bi system. The Bi system composed of the three completely clarifies the relationship and mutual restraint between "What is Bi, why, what to do, and how to do it. This system provides a platform for users and developers engaged in Bi research and practices to share the foundation and communication.
　　
(1) What are the solutions to clear Bi definitions in a three-dimensional architecture?
As a concept, business intelligence was first proposed by Gartner Group, but there is still no universally accepted definition. The current definition includes the concepts and methods for connecting bi as a technology, Bi as a tool, and Bi as a software, as well as Bi, there are many points of view such as automated management processes (see the subsequent article Business Intelligence (BI) three-dimensional architecture-Business Intelligence "operational and improved" conversion ). The author believes that the above points lack systematic thinking and revealing of clear implementation solutions and methodologies, which neither allow users to understand nor guide developers to implement operations.
With the help of the "3D Architecture", I try to fully understand bi. Bi 3D Architecture consists of "3D mode" and "three-layer funnel. Based on the basic principles of the three dimensional model (3D model), the subject, object, and tool are analyzed. The general principles of Bi are clarified through the detailed and comprehensive interpretation of the diagonal color dimension, content dimension, and tool dimension. The three-tier funnel describes the screening process for profit-making organizations in different industries, value links, and development stages to rationally find Bi principles. At the same time, the results of practice will in turn improve the three-dimensional model to improve the general principles of Bi. Figure 5 shows the three-dimensional architecture of Bi.
　　
Figure 5 "3D Architecture" of business intelligence"
　　
In conclusion, business intelligence (BI) is based on the theory of cognition and organization (mainly used for the classification of diagonal colors) and adopts appropriate "Tools ", it aims to help "related roles" make the best decisions on "related content" within the scope of their responsibilities. It is composed of "Three-dimensional model" and "three-tier funnel". It is an overall Auxiliary Decision Support Solution integrating concepts, organization, processes, and technologies of the entire enterprise.
The preceding Bi definition is based on a three-dimensional model and a three-tier funnel. It not only describes the general principle of Bi, which consists of role dimension, tool dimension, and content dimension, in addition, the application scope of Bi is filtered from the industrial/industrial, value chain, and development stages, so as to come down to practice. It can be seen that this definition not only facilitates users who are not familiar with the technology to understand the meaning of Bi, but also allows users and developers to roughly determine the depth and breadth of the needs to be met by using bi for the current and future. Therefore, the Bi definition and its three-dimensional architecture establish a communication platform for users and developers to discuss Bi, develop and use bi.
　　
(2) Three operating modes clarify the classification of Bi results: What kind of Bi should be made
Different enterprises have different business management needs at different stages and can adopt different Bi models to meet them. Bi has three different operating modes for business and technology.
Based on the business and the difficulty and time required by the relevant users to make decisions within the scope of their responsibilities, Bi has three modes of operation: process model, result model and comprehensive model.
The process mode acts on the operation information system, introduces intelligence in the user operation process, provides timely feedback and immediately acts on the workflow, assists the user to make the best decisions for daily work, and improves the work effect.

Result-based mode integrates the intelligence introduced by various operating information systems to generate various presentation results that assist relevant users at different layers to make the best decisions on "matters within the scope of responsibility, such as common reports, dashboards, BCG matrices, and marketing plans. A comprehensive mode is a mode of transformation or critical mode. The transition mode refers to the transition from the process mode to the result mode, or from the result mode to the process mode. The critical mode is applicable to both the result mode and the process mode.
The comprehensive model is suitable for "determining the difficulty and making decisions on the appropriate time-consuming business management issues. Business-Based Bi three modes 6 is shown.
　　
Figure 6 Bi three modes of operation-for business
　　
Based on Technology, Bi has three operational modes based on the degree of data integration: Data Warehouse mode, data mart mode, and line mode. The fundamental difference between the data mart model and the data warehouse model is the product of the top-down and bottom-up construction concepts, that is, the specific embodiment of inmon and Kimball in product application. The parallel mode is a compromise between the two. The data mart model reflects the Bi operation idea of "Quick Start", which can better meet the needs of business users. The data warehouse model focuses on planning and requires enterprise-level data consistency, with a greater emphasis on data quality.
The three business-oriented Bi modes are more suitable for user understanding, and they are also suitable for business-oriented users. It solves the problem of how Bi helps users make the best decisions within the scope of their responsibilities, such as the continuous application of Bi results by employees during system operations, you can still see the analysis results and then use them in other non-information systems such as files and speeches; technical personnel, including users, are more suitable for the data mart mode, data warehouse mode, and line mode.
Based on the three operation modes for technology and application, you can answer users' and developers' actual questions about what kind of Bi to build. Users and developers can build a process-based Bi + data mart based on the actual situation of an enterprise.
(3) four-step development: How to develop bi?
Based on the detailed analysis of Bi tasks, the author summarizes the four-step development methodology (see figure 7 ). The four main links are: 1. model design and technical evaluation; 2. Data collection and data consistency; 3. Front-end display; 4. determine data standards and metadata management.
　　
Figure 7 four-step development methodology