Chapter 2 Source: 6 Origin Processes
1. Four stages of data asset development
1.1 Data Assets 1.0: Building a Consumer Information Database
1.1.1 Initial contact between data side and business side
The first data solution created by the data side for the business side must be successful. The full link content of data items, data processing logic, data usage methods, and data empowerment must be designed for business personnel in the first cooperation. Ensure that business personnel can use data solutions with a low threshold.
Data product managers who are deeply involved in the front line of the business will continue to pass the data requirements of the business department back to the data department, and the data department can build version 0.1 of the data assets based on these data requirements.
1.1.2 Stimulate business personnel’s interest in using data
The data product manager designs some more business-like statistical labels for business personnel, such as "most frequent online time period", "purchasing cycle", etc. These labels are realized through unified function processing and calculation. After business personnel gradually become familiar with tags and their interest in tags is greatly stimulated, the business side will proactively put forward some requirements for algorithmic tags, eg: "predict gender", "predict age" and other basic status attribute tags that need to be predicted through a large number of behaviors . These attribute labels gradually penetrate into the core essence of human beings, and are closer to the business end's requirements for consumer portrait insights. The label information provided on the data side drives and inspires the imagination of data scenarios on the business side.
Only when the value line and the data line resonate at the same frequency and iterate each other can the enterprise's data asset system be truly established. The label team continuously sorts out the available labels with colleagues in the advertising department, formulates a development plan for new labels, completes label production and launches with quality and quantity, and provides the first priority resources to ensure the smooth use of digital resources by the advertising core engine. .
1.2 Data Asset 2.0: ID-Mapping opens up data
- A milestone of ID-Mapping technology in the advertising field is that precision marketing has established a complete closed-loop link from data access → customer identification → crowd selection → perspective analysis → targeted delivery → return optimization; at the same time, this marketing chain The road has the ability to connect with external data resources and share data resources; it demonstrates the feasibility of the advertising alliance ecology from the data level, and has an important role in promoting and impressing the advertising ecosystem.
- Through the continuous verification of data value by consumer tag libraries and the continuous opening up of data between different sectors through ID-Mapping technology, the process of data cooperation and acquisition continues to accelerate.
1.3 Data Asset 3.0: Group-wide data sharing and common prosperity
- Both parties perform their own duties: the data department does a good job in the unified management of data resources and the stable guarantee of call performance, while the business department can focus on business scenario optimization and polishing data innovation references. Data fusion must achieve the effect of 1+1>2. As there are more and more data sources in the consumer tag database, the data volume is getting larger and larger, and the completeness and accuracy of tags are also improving. When the phrase "outside the data center, you can't find a better and more complete data asset than this" becomes a fact, customers will hand over their data backs to you.
- In the data asset 3.0 stage, we are committed to building a data alliance ecosystem.
- The data alliance ecology has formed a stable self-service cycle: each business data system is plugged into an interface for data reflow, and data sources are scheduled periodically and synchronized with the data center to realize data resource updates and automated processing of data assets, and data services configured on demand. It is called and operated smoothly and orderly by various business systems.
1.4 Data Assets 4.0: Data Practices in a Wider Field
After careful consideration, the scene and spatio-temporal dimensions of the tag are increased, so that the tag can truly restore the three-dimensional object in any scene, or the full spectrum information of any slice on the object. In data asset version 4.0, the object concept has been initially refined and expanded, and the thinking and cognition has been improved upstream of the granular processing of tags.
2. Two stages of methodological abstraction
2.1 Methodology 0.1: Overview of methods
2.1.1 Which should we sort out first, labels or objects?
In the methodological process of sorting out data assets, the first thing to do is not to focus directly on the "tag", but to focus on the core essence of the tag - the "object". Only by sorting out and screening out all the objects involved in the business process of a company can the foundation for the growth of the label be established.
2.1.2 What types of objects are there?
Modifications cannot be made on the original concept of "scene" because "scene" and "relationship" are two completely different concepts. Timely synchronize the latest definitions of objects to data product managers, and recommend that they sort out or modify existing data asset designs after understanding the latest object definitions.
2.1.3 Prototype of tag category system
Tags are the carrier of data assets. To achieve a complete description of a certain object, the tags need to be fully sorted out. When the number of tags reaches a certain number, a classification management method for tags is required, that is, a reasonably set tag system. The purpose of labels is to unify standards so that label designers can avoid detours and ensure the quality and effect of data asset design. These normative principles are extracted from various suggestions and requirements put forward by various departments in the whole process of asset design in the past few years, as well as the experience and lessons of the data department itself. They are already a relatively comprehensive collection of labeling norms.
2.2 Methodology 1.0: Principle Research
2.2.1 The importance of basic theory
Only with the basic definition can there be subsequent deduction and derivation, and many disputes and doubts will naturally be resolved. It is important to understand the definition and rationale before taking action.
2.2.2 Theoretical framework based on tree as prototype
The basic principle framework of the tag category system: the basic structure, growth principle, planting and usage mode of the data asset tree. "Tag tree with tree structure" can well deduce the category structure of tag trees, the bionic process of growth and decay, as well as the differences and connections between planting and using trees.
2.2.3 Continuously enrich the complete process
The label category system methodology has officially entered the 1.0 stage, covering basic elements such as origin, principles, methods, and practical implementation, and has managed the applicability examination of a large number of complex projects. The data asset construction model based on tags has been widely disseminated and introduced in leading enterprises in various industries. The optimization iteration of the methodology is interspersed between the bustling customer site, the value-shaping data generation workshop, and the quiet and self-disciplined research desk.
2.3 Positioning of tags in the data system
2.3.1 Position of tags in data assets
Broadly speaking, all data resources owned by an enterprise, including original data, intermediate data, temporary data, data category system, tag category system, tags, tag category system methodology, etc. are data assets.
From a precise definition, data assets refer to data resources owned or controlled by an enterprise that can directly bring economic benefits to the enterprise. Data resources organized in the form of tags are the best way to present data assets. Data assets organized in the form of tags are the best way to present data assets. Since tags are a business-oriented organization method, meta-tag information can make data resources readable and easy to understand; at the same time, the tag-state data organization method is the smallest unit of use and management, making data resources both easy to use and valuable core features. Only by transforming and organizing data resources through tags can we best practice the complete operational link of viewing, selecting, using, managing, and evaluating data assets.
Data assets with tags as the organizational carrier have the following eight significant and unique features:
(1) Ability to confirm rights
All data assets should be cleaned and processed from data sources legally obtained or effectively managed by an enterprise or institution, otherwise they cannot be called assets.
In large group companies, roles with ownership rights, management rights, and usage rights of data assets will be divided: the data source collection and provision department has the ownership rights of data assets; the design, processing, management, and operation departments of data assets have the ownership rights of data assets. Management rights; the use and consumption of data assets The department has the right to use data assets.
(2) Readable
Through data labeling, the hard-to-touch data information is converted into label information that can be obtained in the front-end business, so that the object type can be filtered, the category system can be folded and viewed, and the label list can be read: data personnel or business personnel can call it on demand Basic information such as the category, tag name, tag definition, tag logic, tag value, etc. of any tag. At the same time, in the tag details, you can see the applicable data application scenarios, historical business service calls, and data asset consumption of the tag. Use information such as evaluation feedback from other parties. Tagging makes data readable and promotes the business side's participation in the construction process of digital transformation.
(3) Easy to understand
While tagging data, meta tags are used to convert difficult-to-understand data terms into popular business terms, and meta tag information is fully recorded through standardized operations during tag creation and design. The transformation from data to labels not only realizes the conversion and mapping of data terms to business terms, but also in the label design process, the label methodology requires each label designer to fill in the "Label Detailed Design Document" in accordance with the specifications.
(4) Easy to use
Tagged data assets can be used to create data units cut into the smallest granularity. The idea of using tags is also closer to the business end: encapsulating the smallest reusable unit of data into a "commodity". Tags innovate a data usage model: break down data into the smallest granularity unit, and flexibly select the required parts by building blocks each time, and quickly complete the assembly of a certain data service or data application through tools or platform support .
(5) Measurable
The search volume, page view volume, application volume, call volume, etc. of a certain tag data can be recorded and measured by the system. Measurable characteristics are conducive to the optimization and operation of tags, help control the safe use of tags, and evaluate the business use value of tags.
(6) There is a price
There must be a measurable value ruling for data assets. In the business world of data value exploration, attention must be paid to cost support and profit returns; data assets have costs such as collection, production, management, and operation, and users of data assets need to "book" or "pay" for the use of data assets. At the same time, data managers must continue to optimize and update the best configuration of data assets from the consideration of value.
(7) Controllable
Data assets must be controllable, otherwise there will be huge security risks and management costs. Tagged data assets can be managed through the tag management system for full life cycle operations, including meta tag information management, tag standard management, tag security management, tag quality management, tag cost management, tag value management, etc.
(8) Value-added
Data assets are special assets that become more and more used. As long as data assets are built in accordance with standard action specifications and operated with data value as the guide, their value will continue to iterate and have unlimited value-added space.
2.3.2 The position of the label in the data
(1) What is a data center?
- From an architectural point of view, the data center is a link between the previous and the next data accumulation. Through its own data platform tools, the original data is processed into data assets, and data application scenarios are launched through the service of data assets to help the business or management end reduce costs and increase efficiency.
- From the perspective of implementation, the data middle platform takes data assets as the core, and takes the realization of a series of goals that data assets are visible, understandable, usable, and operable as the starting point, and is equipped with necessary links such as platform tools, process specifications, and application construction, and finally lands data solutions.
(2) The important position of labels in it
- The data center is located between the cloud base and upper-layer business applications, that is, between the stable and heavy technical backend and the flexible and changeable business frontend. Through the data center's abstract encapsulation of underlying complex technical capabilities, front-end businesses can use data capabilities freely and easily, bridging the problem of inconsistent pace between front-end and back-end.
- Within the data center, it is specifically divided into modules such as development tool layer, data asset layer, asset management layer, data service layer, data operation system, and data security system. The original imported data is transformed into the enterprise's own data assets through the development tool layer; the data assets are continuously managed and optimized at the asset management level; finally, the data assets are transported to all ends of the business through asset servitization to realize data value; unification The operating system and standard security management mainly ensure the smooth and orderly operation of the entire data center from the process mechanism level. In the data center, development or management tools can be purchased directly, and the methodologies of operating systems and standard security specifications can be learned. However, data assets and data services must be the results of the company's own construction and implementation. They are the core of the data center. At its core, there are no shortcuts.
- The core of data assets and data services is tags: data asset identities use tags as organizational carriers, and data services are essentially a value pipeline that delivers tags to the business end for use. Tags are the “core core” of the data center’s value chain.
2.4 Definition and explanation of key terms
(1) Data
Data refers to the symbols that record and identify objective events, and are physical symbols or a combination of these physical symbols that record the nature, state, and relationship of objective things.
(2) Data assets
Data resources owned or controlled by an enterprise that can directly bring economic benefits to the enterprise.
(3) Data center
The data center is a set of mechanisms that can sustainably “make enterprise data available”.
(4) Label
Tags refer to data carriers that are processed from original data and can be directly used by businesses and generate business value.
(5) Meta tag
The meta tag is the tag of the tag, which is to sort out the attribute information of the tag (especially the business attribute information).
(6) Category system
Category system refers to the classification, structure, and organization method of a certain type of items (things).
(7) Data category system
The data category system is a directory structure formed by sorting out the original data fields owned by the enterprise using a category system.
(8) Label category system
The tag category system is a directory structure formed by sorting out the tags required for enterprise business using a category system.
(9) Object
Objectives to be studied in the real world.
(10 people
An object that actively initiates behavioral actions.
(11) Things
Passive objects in behavioral actions.
(12) Relationship
Some kind of connection between two objects, such as people and things, any person, things and things.
(13) Scene
The performance of specific objects (people, objects, relationships) in time and space under a certain environment.
(14)Backend category system
The backend category system is oriented to data asset managers. It is a complete set of enterprise data assets and is relatively stable. Tags are mounted, viewed, and managed according to a unified classification method.
(15) Front desk category system
According to the needs of the scene, the tags can be organized into new categories according to the front-end scene to form a front-end category system.