Business Intelligence/Frame Architecture

Purpose:
Input: Framework architecture and scorecards
Activities: Diagram frame architecture (Legorreta and Valentine)
Outputs: Frame architecture - conceptual, logical and physical
Documents: Frame Architecture

A frame turns data into information that is useful to the decision maker. It is therefore important to first define the terms related to data and information.

Hierarchy of meaning: From data to events

Data has a hierarchy of subjective meaning. Through Extracting, Transforming, Loading, Cubing, and Analyzing techniques:

Data becomes information
Becomes knowledge
Becomes intelligence
Becomes events

Data: Data is objective. For instance, data on a car can be color, cost, size of the engine, etc.

Information: Information is subjective as it depends on the meaning the subject gives the data. Meaning is acquired by adding structure and/or context to data. Context can be added either explicitly (metadata)or tacitly by the subject.

Knowledge: Knowledge is information that fits a reference frame. A reference frame is a model the subject uses to comprehend reality. Knowledge is subjective to the extent that the reference frame is unique to the subject.

Intelligence: Intelligence is actionable knowledge. Intelligence is subjective because different subjects will have different reasons prompting them to action even when they share many reference frames. They share the same “know-how and know-what ”frames but not the same “know-why ” frame. Hence the need for leadership to prompt a timely unified response. Intelligence creates cognitive arousal. However, intelligence need not trigger a response.

Event: Event is intelligence that demands a decision from the subject. What is an Event is subjective because different subjects will hold different motivations prompting them to action even when they share the same intelligence. They share the same “know-how, know-what, and know-why ”frames but not the same level of commitment.

Hence the need for leadership to prompt a timely, committed, unified response. Events may create emotional arousal in addition to cognitive arousal. When this happens we say the event is provocative. Emotional arousal may, at times, override cognitive arousal. “Do nothing ”is often an option in the menu of decision choices. Even if the event is very provocative, the subject may decide to do nothing. Events add up beyond a threshold of latent summation to trigger Action. The threshold of latent summation can be visualized as a bar that fills up –like when you are waiting for something on a computer; except here you will see stops, reversals, and hesitations, before the bar fills up. Action yields results, which is information regarding the apparent consequents of the action.

The information architecture identifies how data moves to events and action. The information architecture identifies a process for adding structure to data so as to move it through the hierarchy of meaning.

ETLCZ

Information Architecture is the pattern of activities data undergoes as it flows up the BI Frames (i.e., through the hierarchy of meaning). An Information Architecture can be depicted using letters: E,T,L,C,Z:

E = Extracting
T =Transforming
L =Loading
C =Cubing (short for “slicing and dicing ”)
Z = Analyzing Extracting needs to include IQA

The information architecture follows some rules:

E starts the Information Architecture (1)
T is optional, but can occur many times (0 …)
L is not optional and can occur many times (1 …)
C is optional and can occur at most once preceding Z (01)
Z ends the Information Architecture (1)

Some examples of Information Architecture: ETLZ, ETLTLCZ, and ELTCZ

ELZ, the simplest Information Architecture, is type-0 E(n-times TL)C*Z is type T-first E(n-times LT)C*Z is type L-first

Business Intelligence Framework

Conceptual schemas depict frames as abstractions
Logical schemas add logical relationships between the constructs in a conceptual schema in increasing levels of detail
Physical schemas are real-world instances of the logical schemas and include design specifications in increasing levels of detail