When building a BI-solution or other structured database some challenging questions have to be answered:
Ø How are the knowledge workers in your organization effectively involved in the design of the database systems that produce the information?
Ø How do you set up a correct and unambiguously defined information product specification?
Ø How do you maintain this information product specification effectively and efficiently?
Ø How does a methodical approach contribute to answering these questions?
The answer to the fourth question is simple. To create an information system that produces high-quality information consistently our approach must guarantee a high-quality information product specification that can be employed and maintained efficiently. In doing so, the approach must ensure that the two root causes for failure, mentioned on our home page, are fully eliminated.
To this end, our Information Management Frame approach is fully communication orientated. We do not model reality itself, but rather the communication of the business experts about their reality, and this in terms of true fact expressions. While doing this, we also ensure that we model all conceptual aspects of the communication, nothing more and nothing less. This makes our approach efficacious. Secondly, we ensure that our domain experts are able to validate completely that their communication has been modeled correctly, without referring to any for them unintelligible IT-language, pictures, or technology aspects. This makes our approach effective. Finally, we capture all relevant metadata generated in this process in a single repository, from which the metadata are replicated or transformed to any place desired by means of fully automated algorithms. This makes our approach efficient. Thus we have met the three 'E-criteria' for judging a method in Soft Systems Methodology of Peter Checkland.
The Information Management Frame (IMF), shown in the figures under Impressions top right, consists of layers and columns. Following the 1975 ANSI standards for data models in data intensive domains, the layers are the 'Design layer' and the 'Implementation layer'. The 'Design layer' consists of two levels. On top it has the Conceptual level, which focuses on the domain area, the Universe of Discourse of the business experts. This is the only level where we can meet the information customers. The second level is the Logical level, which focuses on data structures: objects, relationships, and properties. This is the top level of all ER-tools. The 'Implementation layer' also consists of two levels. The first one is the Physical level, which focuses on the physical platform used to store the data. This is the Relational Model level, the bottom level of all ER-tools. The second level is the Technical level, which focuses on the practical use of the physical platform chosen.
As columns in our IMF we have several 'pillars' of application. The figures top right show the Information Management Frame for Business Intelligence projects. The pillars can be adapted to any kind of project, but the pillar 'Corporate Fact Base' must always be there, because it is common to all applications involving the integration of information over different enterprise processes. This pillar is meant for an enterprise-wide data model that is devoid of ambiguities in the communication between the various company processes. A full explanation of the IMF and its use in projects is given in our courses. Prominent in our IMF approach is the use of Fact Oriented Modeling.