Академический Документы
Профессиональный Документы
Культура Документы
Setting modelling guidelines that analysts can refer to while they are producing models, 2. Setting guidelines that reviewers can use when they are evaluating models during an approval or delivery stage in the project 3. And making sure that the project can deal with the inevitable need for change and refinement in guidelines
1.
The following sections look at each of these points of view in more detail.
5. 6.
Producing Models
One of the key issues to address is how to allocate modelling responsibility between individuals in a team of analysts. Any project will proceed from the start with some system of dividing up the modelling work. For example, analysts may each be assigned responsibility for modelling business processes within a given business unit or subject matter domain. However, teams of modellers within a given modelling project always have to find ways to share some entities, concepts, or models. If the team is using a business process modelling software tool, then there will be issues regarding how to physically share data objects. If not, there are still new identifiers, concepts, or terms that must be used in common. Experienced analysts will recognise that communication of knowledge and information between team members is a key factor influencing how quickly a project can progress. The pratical aim of modelling guidelines should be to assist good communication as well as avoiding error.
Roles
We recommend that projects must clearly define the roles of modellers, including the distinct responsibilities for each type of modellers. A key aspect of this role definition is defining ownership for distinct sets of models and entities such as processes, activities, requirements. The user accounts that are created for software modelling tools should be logically related to the roles of the users. The modelling guidelines may be the appropriate place to document user accounts, groups, permissions, and how these relate to the modelling tasks.
Communication
Projects and teams need to develop good processes and habits of communication, although this is not necessarily controlled by guidelines. Policies about how often teams meet, and what kinds of modelling issues require email notification might be mentioned in guidelines. Communications with key stakeholders or external stakeholders is more formal and may be covered in modelling guidelines because models may form an important part of stakeholder communications.
Sharing Data
Even when using software modelling tools, the ability of a team to share access while avoid conflicting changes depends as much on the discipline of the users as the functions of the system. We recommend documenting the steps that team members follow for all common data access steps.
Security
We recommend that guidelines should describe the security objectives of the project in practical terms as well as specifying rules that analysts must follow. As much as possible correct setting of permissions on assets should be an automatic consequence of following storing and sharing processes.
Publishing
Publishing models from an on-line repository is often requires detailed or complex steps. This is particularly true if there are numerous publishing formats or non-default formats required. We recommend that publishing steps for modellers be thoroughly documented. Responsibility for publishing should be delegated to a special publishing role as much as possible.
Prescriptive Rules
It should be clear that some of the rules are requirements of the project, and these requirements should be written so that a modeller can quickly assess whether they have been met.
2.
Non-prescriptive Rules
The guidelines should include rules that are only helpful hints. These should cover the most problematic common issues that modellers face. Aspects of quality that are subjective or difficult to evaluate should be treated as non-prescriptive for practical reasons.
3.
We recommend that the standards and recommendations that modellers are meant to follow be documented as a checklist of prescriptive rules, plus a manual of non-prescriptive rules. This same checklist should also serve as the basis for evaluation during the peer review process.
Projects very often take an incremental approach, and reviews of various scope may be scheduled at any point in a project. As we noted above, the guidelines used for evaluation are based on a base of standards and recommendations that would also be built into the modelling guidelines. We recommend that the standards and recommendations that modellers are meant to follow be documented as a checklist of prescriptive rules, plus a manual of non-prescriptive rules. This same checklist should also serve as the basis for evaluation during the peer review process.
Roles
A good modeller should also be a good reviewer. A business modeller will naturally be most interested in feedback from a skilled and knowledgeable business analyst. We recommend that reviewing skills and responsibilities be developed and required for roles that involve modelling skills and responsibilities.
Communication
We recommend that the process for scheduling peer reviews and providing feedback be documented.
Reporting
The requirements for delivering models for approval or finalisation, such as format of reports, should be established or discovered ahead of time so that the processes and responsibility for preparing reports can be properly considered and documented as part of the evaluation guidelines.
Acceptability or compatibility of publishing format Time required to publish Requirements tracing (can models be mapped to project requirements) Inaccuracy or error (are there too many errors in the models) Legacy import (do the new models need to integrate with legacy models) Tool support and skill set (does the team have the skills to use or administer the needed tools) 7. Security requirements 8. Data modelling (can system implementers extract the necessary data model from the business models) 9. Knowledge capture (are modellers creating ad hoc documentation because model doesnt accommodate all the concepts they want to capture)
1. 2. 3. 4. 5. 6.