Information and Systems

BA 370 - Information Systems

Information:

Living organisms, some machines and aggregated types; e.g., organizations need information to function properly.

Example: trial & error:

Eat raw potato --> not so good --> cook potato first --> much better.

All learning requires information: facts, theories, rules of inference, verification, falsification (K.R. Popper: 1902 - 1994).

Problem: Why do organisms learn anyway?

Information enables learning through the application of memory and logic (induction and deduction).

Information is distinguished from data through relevance.

Problem: What do you think about Stair & Reynolds' definition of Information (p.5).

Relevance as a logical construct:

Tautologies: e.g., All crows are black or not black.

Redundancies:

Capacity
Coal
Cooling
Develop

1
Large
No
Water
Yes

2
Medium
Yes
Water
No

3
Medium
No
Air
Yes

4
Large
No
Air
Yes

5
Small
No
Air
No

6
Small
No
Water
No

7
Medium
No
Water
Yes

8
Small
Yes
Air
No

9
Medium
Yes
Air
No

10
Large
Yes
Air
Yes

Capacity	Small			Medium				Large
Cooling	Water	Air		Water		Air		Water	Air
Coal	No	Yes	No	Yes	No	Yes	No	No	Yes	No

Develop	No	No	No	No	Yes	No	Yes	Yes	Yes	Yes
	1	2	3	4	5	6	7	8	9	10

Capacity
Small
Medium
Large

Cooling
-
Water
Air
-

Coal
-
Yes
No
Yes
No
-

Develop
No
No
Yes
No
Yes
Yes

1
2
3
4
5
6

Capacity
Small
Medium
Large

Coal
-
Yes
No
-

Develop
No
No
Yes
Yes

1
2
3
4

Compiler optimization.

Relevance as an empirical construct: Information reduces future uncertainty (Stair & Reynolds, p. 7).

Problem: Can you spot the information (relevant data)? Monty Hall Puzzle:

Information may not be accurate:

Precision.

Reliability.

Information should be checked: Quality Assurance (QA) and Quality Control (QC).

Information has a time dimension.

Immaterial; can be stored on computers; nonphysical; does not reduce.

It's everywhere, but costly to get. Hence, it's a commodity and has value.

Data come in various types:

Stair & Reynolds (p. 6).

Quantifiable types (meansurement theory):

Nominal

Ordinal

Interval

Ratio

We have much more than we think: let's go find it (BI and data mining)!

By discrete pattern recognition:

1, 2, 3, 5, 8, 13, 21, ?

23, 5, 12, 3, 15, 13, ?

100, 365, 24, 60, ?

By trend pattern recognition: e.g., Linear regression.

Overall: The many sides of information: Stair and Reynolds (Table 1.2, p. 8)

Formal: Information Theory (Shannon): Gentle primer by Thomas Schneider:

Information: a decrease in uncertainty

Uncertainty of information: Average number of bits (1's and 0's) required to represent a symbol (signal) in a string of symbols (message):

H = -ΣP_ilog₂P_i

where:

P_i: (prior) probability of symbol i.

-log₂P_i: surprisal of symbol i (study the nature of this relationship with a chart in a spreadsheet!).

Example:

Let string of symbols: ACATGAACAACTAGT... (note that this string can be infinitely long).

Let P(A) = 1/2; P(C) = 2/8; P(T) = P(G) = 1/8.

Note: P(symbol) ≠ symbol count / number of symbols in the string !!!!

Surprisal(A) = 1; Surprisal(C) = 2; Surprisal(G) = Surprisal(T) = 3.

H(ACATGAACAACTAGT...) = 1/2 * 1 + 2/8 * 2 + 1/8 * 3 + 1/8 * 3 = 1 6/8 = 1.75 'bits'

H is an (expected) average (assumes Shannon-Fano Coding)!!

AAA = 111; 3 bits / 3 symbols = 1 bit / symbol.

CCC = 010101; 6 bits / 3 symbols = 2 bits / symbol.

TGT = 001000001: 9 bits / 3 symbols = 3 bits /symbol.

ACA = 1011; 4 bits / 3 symbols = 1.334 bits / symbol.

AAAACCGT = 11110101000001 = 14 bits / 8 symbols = 1.75 bits / symbol.

Problem: Why do we use 01 (C) but not 00? After all, we do use 001 (T) and 000 (G).

Some questions:

What's H if there's only one symbol?

What's the maximum value for H and what does that value signify?

What's the minimum value of H and what does it signify?

Which has a higher H? All P(X)s equal or not all P(X)s equal? Why? (See Schneider figure p. 4).
Which has a higher H? All P(X)s equal or all P(Y)s equal when P(X) > P(Y)? Why?

Information = decrease in uncertainty: R = H_before - H_after

If H_before = 1.75 bits/symbol and H_after = 1 bit symbol; you gained (got information) of .75 bits/symbol.

Applications: data compressions, noise modeling (see Shannon's / Schneider's example on p. 7).

Problem: Recall the Monty Hall puzzle. We concluded that by opening one of the remaining doors, you gained information. Can you use Shannon's equations to compute how much information you were given?

Information Systems:

Heylighen et al.: "...however complex or diverse the world that we experience, its organization can be described by concepts and principles which are independent from the specific domain at which we are looking."

A system is a conceptualization of reality. System is a cognitive (human) construct.

Elements.

Relationships.

Structure.

Systems are dynamic; i.e., they have behavior.

State.

State transitions: S _t+1 = S_t + B

State space.

Environment and boundary.

Inputs transfer messages (information or energy) from the environment or from elements to other elements.

Outputs transfer messages (information or energy) from the elements to other elements or to the environment.

Open systems accept input from the environment; closed systems do not.

Systems have no inherent boundaries and hence are not inherently open or closed (Hence, Heylighen's insistence on domain independence).

Systems which change their behavior patterns as a consequence of their openness are adaptive systems.

Feedback: an input to an element as a response to a previous output by the same element.

control elements: elements whose input is under control of a decision maker.

An element can be a system by itself (subsystem).

An element modeled as if it has no structure (elements & relationships) is a black box.

For these and other classifications, see Stair & Reynolds, Table 1.3, p. 10).

Problem: Can you model (Stair & Reynolds, p. 13) a standard heating system consisting of a thermostat (itself consisting of a heat-sensing device such as a thermistor, a set point and a switch), a furnace, rooms, raditors or air ducts, etc.? Does the system have feedbacks? Does it have control elements?

Second Law of Thermodynamics: The entropy {S=kLnW} of a closed system never decreases and increases whenever possible ( the entropy page).

To maintain or decrease entropy, a system must use energy; i.e., increase entropy elsewhere.

Problem: does the concept of entropy apply to information systems? How?

Information Systems:

O'Brien (2001): "a group of interrelated components working together towards a common goal by accepting inputs and producing outputs in an organized transformation process." (p. 40).

Alliance for Telecommunications Industry Solutions (ATIS):

"A system, whether automated or manual, that comprises people, machines, and/or methods organized to collect, process, transmit, and disseminate data that represent user information."

"Any telecommunications and/or computer related equipment or interconnected system or subsystems of equipment that is used in the acquisition, storage, manipulation, management, movement, control, display, switching, interchange, transmission, or reception of voice and/or data, and includes software, firmware, and hardware."

"The entire infrastructure, organization, personnel, and components for the collection, processing, storage, transmission, display, dissemination, and disposition of information."

Jessup/Valachic (2003): "Information systems are combinations of hardware, software and telecommunications networks that people build and use to collect, create and distribute useful data, typically in organizational settings."

Stair & Reynolds (p. 15): "a set of interrelated elements or components that collect (input), manipulate (process) and store, and disseminate (output) data and information and provide a feedback mechanism to meet an objective."

Problem: Can you do better?

Problem: So what are the skills/characteristics of a good MISer?