DMN introduction

Recommended for you: Get network issues from WhatsUp Gold. Not end users.

Background

The purpose of DMN is to provide a model of decision structure, so that the organization's strategies can be depicted graphically clear, through the analysis of the business of the accurate definition, its automation (optional).

This decision applications discussed mainly for the following two different view of modeling standard:

However, some developers (including the submission of group members) found that modeling standard for more than two of decision making, the internal structure of its own is not convenient. DMN will provide third perspectives: "the decision demand map" to form a bridge between business process models and decision logic model:

Together, the decision-making and decision logic can provide a complete decision model, decision process through the specified tasks, do a very good complement for the business process model. These three aspects of the relationship between models as shown in Figure 1.

The results will be the business process the business rules and the role of analysis model, model of cross validation, "TOP-DOWN" process design and automation and decision automatically execute a series of detailed modeling together. (BPMS call through a deployment in BRMS decision-making service)

Although Figure 1 shows a business process model and decision model of the contact, the purpose is to show the relationship between DMN and other standards, but it must be stressed that DMN does not depend on BPMN, they are two aspects of the decision demand and decision logic: it can be used alone or with the use of the domain model, decision there wasn't any references to business process.

DMN provides modeling structure decision demand and decision logic:

Again, although the figure 2 depicts the decision modeling structure are interrelated, but it still can be used separately or used in any combination. For example, you can use DMN only DRDs, or only the definition of decision table, or just write FEEL expressions.

The basic concept

The decision-making layer

"The decision of the word "usually has two kinds of definition: 1 according to the given condition from the candidate selection process in a certain condition is selected; 2. In this specification, we use the former usage, aPolicy decisionIs: from multipleInputValue judgment in aThe output value(has been chosen conditions) behavior, through the definition of logic is how to determine the output from the input. Decision logicMay contain one or moreBusiness knowledge model, The business rules, analysis model, professional knowledge or other form of packaging business. This is the basic structure to construct the decision model of all, shown in Figure 3.

The official can define decision or business knowledge model, which may be (for example) the domain experts responsible for the definition and maintenance, or from the business knowledge model derived from the source document, or is the test case set, they must be consistent with the decision. These are called knowledge sources (see Figure 4).


A decision is also called the "dependent", it is used to decide the input output it. The input can beInput data, Or other decision output. (they can be in these two cases is the data structure, rather than simple data item. "Decision 1") if the input contains "decision output item 2", 1 "dependent" decision 2 decision. The decision may therefore be connected into a network, this is called decision demand diagram (DRG), it can be drawn as a decision-making demand diagram(DRD).

A DRD display a set of decision how to depend on each other, on the input data, and the knowledge of the business model. As shown in Figure 5, one of only 2 decision DRD simple example:

A decision may have more than one business knowledge model requires, a knowledge of the business model and many other business knowledge model, as shown in figure 6. This will allow complex decision logic of knowledge through the combination of different fields can be modeled, and provides the decision logic alternative versions in different conditions.

 

DRGs and them as symbols of DRDs, detailed in the next chapter.

Decision logic layer

The decision-making layer decision model as a component can be described, they are macroscopic, only for the service concept. Describe the hierarchy is usually analyzed fully oriented decision-making in the areas of business, to identify the business decisions, their mutual relations, business knowledge and the required data, and business knowledge source. The use of decision logic, the same components can be described in greater detail, to obtain a complete set of business rules and calculation for, and (if needed) allows decision fully automated.

Decision logic also provides some additional information, such as on how to display elements in the model. For example, a decision table decision logic elements can specify whether to display the row or column rules. For the decision logic element calculation can specify arranged in vertical or horizontal.

The corresponding relationship between the concept of decision-making demand level and decision logic layer will be described in later. Please note, in the following diagram, as shown in Figure 1 and Figure 2, gray oval and dotted line to draw only for the purpose show a correspondence between the concepts of different levels. They are not a part of DMN symbols, they will be defined in the DRG and DRD symbols, symbols and FEEL decision table notation. It is envisaged, the implementation will provide the ability to move between the model hierarchy, such as "open", "drill down" or "amplification", but DMN does not specify how this should be done.

In the decision layer, the DRG in each decision is defined as the use ofValue expression, It specifies the decision is how to determine the output input. In this level, the decision is considered to be the execution of expressions. Value expressions can use boxed expressions to sign, as shown in Figure 7.

In the same way, in the decision-making logic layer, the definition of a business model is the use of a value expression, it specifies an output is computed by a group of input. Value expressions can be encapsulated intoFunction, From the value expression calls in the decision; business knowledge model is the function example (but the decision logic may also contain business knowledge model does not have the function). Explain the business knowledge model like the DMN function, as shown in Figure 6 combined knowledge of the business model, with clear functional combination of semantic. Knowledge of the business model of value expressions you can use the box function to mark, as shown in Figure 8.

Multiple decision logic in a business knowledge model can be included, can be expressed as a function. This will enable many existing decision logic modeling standards (for example, business rules and analysis model) can be imported into DMN. Decision table is an important form of business knowledge, explicit support in DMN. The knowledge model of this kind of business can use decision table flag, as shown in Figure 9.

In most cases, the decision logic is encapsulated into business knowledge model, then the business knowledge model is how to call the decision is related to the value of the expression, the result is output is calculated to call the combination. The decision value expressions may be entirely through their own specifies how to decide the output from its input, without invoking the business knowledge model: in this case, no business knowledge model and decision making (whether in the decision-making layer or in the decision-making layer).

In DMN, a decision logic is the use of the expression language to define, describe in detail the FEEL expression language described in Chapter.

FEEL: the Friendly Enough Expression Language

The range of DMN and Application

As a business or organization in the understanding and definition of decision-making is to design the decision model through the analysis of the business. This decision is the general operation decision for day-to-day business processes, rather than fewer rules or strategic decision facts.

The following three DMN usage:

  1. Artificial decision modeling
  2. Automated decision requirement modeling
  3. Automated decision

Artificial decision modeling

People in organizations can use DMN to model the decision. Artificial decision can be decomposed into a network of interdependent decision making, model using DRD. The DRD's decision may be used to describe the natural language, not a decision logic.

The source of knowledge can be defined for: people (managers), regulators (such as the inspector), file (for example, a policy file) or legislation (e.g., government regulations) management decision model. These knowledge sources can be connected, for example, a decision is made by: the strategy document maintenance provides a set of the definition of the regulatory agencies and the manager of the company.

Knowledge of the business model can be used to extract business knowledge areas in decision making. This will allow DMN to be used for knowledge management need to formally define requirements for tools. Knowledge of the business model can be linked, to show the interdependency between knowledge (similar to the way the use of Knowledge Structure Mapping). Knowledge sources can be related to the business knowledge model is used to represent business how knowledge management and maintenance, for example, documents of the company strategy (source) defines a group of business strategy (business model).

In some cases, for the establishment of a decision needs to be possible to define the specific rules or algorithms. They can use the business knowledge model in DRD logic modeling decisions (e.g., business rules or decision tables), both description form (record decision at present is how to do, or to a particular time how they do) or standard form (define decisions should how to perform, or execute) will be in the future.

Decision models in DMN can be mapped to the use of BPMN modeling business process tasks or activities. The macroscopic point of view, cooperation decision task can be mapped to a set of DRD in a subset of decision making, it represents a group or department's overall decision behavior. From the point of detail, he can build a dependency model in decision using a single BPMN collaborative or group: each participant in the decision by independent collaboration diagram pool (pool) to say, in the model is independent of the DRD. The DRDs decision mapping to the pool of tasks, the input data in DRD is mapped to the contents of the message, transfer between multiple pool.

The combination of BPMN and DMN provides a graphical language, used to describe the artificial decision within an organization hierarchy, detailed definition from the business process activities to the decision logic. In this context, DMN model will describe the collaboration decision, management, as well as their business knowledge dependency.

Automated decision requirement modeling

The use of DMN for automated decision making requirement modeling is similar to artificial decision, but it is entirely command form, rather than describing form, and there is more emphasis on specific decision logic.

Decision making automation, decision logic must be complete, which can provide the decision results for any possible value of the input data.

However, part of the automation is relatively common, some decision is still manually. Modeling of collaborative interaction between manual and automated decision making can be used for the above mentioned, manual and automatic decision machine are respectively provided independent pool, or more simple allocation decision to the business process model of independent tasks: manual decision using automatic decision task, using business rules. So, for example, an automated business rules task may decide to certain things to artificial audit; decision logic automation tasks need to specify the complete, but the audit decision-making can not specified.

Once a DRD decision is mapped to a BPMN flow chart of the task, he will be able to verify through the two level model. For example, it can be verified in DRDs all of the input data is provided by the previous tasks, and the business process in the follow-up to the task and the gateway using the results of the decision making. The DMN model of the relationship between decision making and business process, which can be identified on the business process must be completed to make decisions, so you can specify a specific decision task execution decision. In DMN1.0 do not provide formal mapping from data object DMN ItemDefinitions or DMN InputData to BPMN, but may include examination in some cases mapping.

In conclusion, BPMN and DMN can automate decision-making needs of standardization, and interaction in the business process in it and artificial decision. These requirements can be at any level of detail to specify, or at all levels. The business process model, the three layer mapping between DRDs and decision logic will allow the definition based on computer aided design tools to support these requirements model.

Automated decision

If all decision-making and business knowledge model makes full use of logical decision rules, so it can implement the decision model.

One possibility is that the BPMS (Business Process Management System) is called a deployment in BRMS (Business Rules Management System) in the "decision service". A decision support service package DRD decision logic, provides corresponding to the DRD input data and decision making interface. When the call and a set of input data, making the service will perform the specified decision and return their results. DMN requires that all decision making logic is no side effect, which means that the decision of service will be in accordance with the principles of SOA, simplify system design.

A decision model of the structure is visible in the DRD, can be used as a basis for planning and project implementation. Project specific tasks include: define the decision logic (for example, expert rule induction or creating analysis model and decision model of component).

Decision logic is encapsulated in a decision: in the service business knowledge needed by hand for long-term maintenance, and the use of a special "knowledge maintenance interface" decision making. DMN supports effectively design and implement the knowledge maintenance interface: any need to maintain knowledge of the business should establish business knowledge model and knowledge sources in DRD. DRD is required for maintenance of knowledge and the user interface provides a standard, a decision logic also designated business knowledge initialization configuration, easy maintenance.

Other update decision logic needed by conventional analysis modeling. As knowledge of the business model in DMN is a function, very simple to use its analysis model in the decision-making service: any one analysis model can express as a function, into the decision-making service directly after the call.

Combined application of modeling

The three cases are not mutually exclusive; a large processing automation projects may also use three ways DMN.

First of all, in the existing treatment decisions can be modeled, to determine the current decision and relates to the full range of business knowledge. Provide process improvement baseline this "AS-IS" analysis.

Next, the process may be re design, to effectively use the automatic and manual decision, usually by collaboration between the two (for example, the use of automatic calculation of artificial decision makers, or the use of decision support system recommended or limit the users). Including such redesign: modeling in each individual task, organization or a set of roles and responsibilities of the decision demand. The model provides the required process and the coordinates of the "TO-BE" standard.

Comparison of "AS-IS" and "TO-BE" model will find the demand not merely automation technology, but on the management of change: the change relates to the roles and responsibilities as well as new or adjust the business knowledge training support.

Finally, the "TO-BE" model will be implemented as software executable. Decision logic is provided with FEEL or other extended logic (Java externally defined or PMML models), component model can be directly realized by software.

DMN does not provide any specific methods for the implementation of the above activities, it only supports the use of their model.

Recommended from our users: Dynamic Network Monitoring from WhatsUp Gold from IPSwitch. Free Download

Posted by Gale at August 08, 2014 - 6:24 AM