Week 3 - Data, information, and knowledge

2/2/00 abstract
Shannon, C.E., and Weaver, W. (1963/1949). The mathematical theory of communication (p. 31-35 only). Urbana: University of Illinois Press.

Presents a model of communication consisting of an information source, transmitter, channel, receiver, and destination.

1/26/00 abstract
Wersig, G. (1997). Information theory. In J. Feather and P. Sturges (Eds.), International encyclopedia of information and library science (p. 220-223 only). London: Routledge.

Reviews the various meanings proscribed to ``information'' and ``information theory''. Shannon's view of information as relating to the entropy of a message is still widely accepted, but various disciplines have introduced competing definitions for ``information'', including: the structures of the world, knowledge developed on the basis of perception, the meaning assigned to a message, the process of communicating knowledge, and the effect produced by communication.

1/26/00 abstract and discussion
Reddy, M.J. (1979). The conduit metaphor - a case of frame conflict in our language about language. In A. Ortony (Ed.), Metaphor and thought (p. 284-297 only). Cambridge: Cambridge University Press.

Argues for a new view of communication as inherently introducing entropy. English affects the way people think about language, by encouraging them to use metaphors that imply words and sentences contain thoughts and ideas. This ``conduit metaphor'' causes people to focus on the speaker as the source of problems in communication. By looking at communication from ``the toolmaker's paradigm'', it can be determined that language doesn't contain as much information as generally assumed, and the receiver of communication can add as much error to the process as the sender.

verbalization	$\rightarrow$ conduit $\rightarrow$	heard
$\uparrow$		$\downarrow$
original		received
thought		thought

In their Conceptual Dependency language, Schank and Abelson called this process MTRANS (Mental TRANSfer). In an MTRANS, a piece of information is transferred whole from one actor to another. The medium (conduit) is irrelevant.

In reality, information can't be transferred verbatim from one person to another, since all information is dependent on context. In order to accurately transfer a single piece of information, the entire contents of the person's mind would need to be transferred as well. Even then, the conduit may have some effect on the transfer.

1/26/00 abstract
Buckland, M. (1991). Information as thing. JASIS, 42, 351-360.

Argues that physical types of information are equally important as knowledge and process types of information. Information-as-thing is the only type of information that information systems can deal with. The information value of an object is situational, so any object may be considered information in some situation. Some objects, like events, are impossible to store in an information system, so a representation must be used.

1/26/00 abstract
Tesar, P. (1991). The other side of types. In G. Rockcastle (Ed.) Midgård Monographs of Architectural Theory and Criticism, Number 2 (p. 165-175).

People automatically classify objects at some level that is useful to them. Intelligent creatures desire to maximize the predictability of the environment with the least effort expended. However, relevant features depend on the bias of the classifier. A new object is experienced in comparison to other objects of that natural type that have been previously experienced.

1/26/00 assignment

What is Information? Knowledge, of Course!

Information can be viewed from a variety of viewpoints, and the viewpoint that is most useful depends on the application. Information scientists would do well to consider information-as-thing, since that is what they have the most control over. Teachers, who deal with the transmission of knowledge, should consider information-as-process. However, information-as-knowledge is the way most people see the world.

Information as Thing

Buckland (1991) argues for the viewpoint of information-as-thing. This view is particularly useful for building information systems, since a system has no access to knowledge or process information. However, this view is problematic, because text, the information most accessible to information systems, is a secondary form of information. At best, it is the product of a person trying to express the information (knowledge) in their mind. It could be much farther removed from the original source, as is the information in a news article. A reporter sees an event happening, writes down some notes, later reads the notes, and composes an article. The information system, having access only to the article, has lost much of the nuance of the actual event, and added the bias of the reporter.

Systems that have direct access to the world can gain their information from the world, but few systems can sense the world directly, and they can only access limited types of information.

Information as Process

In this view, all objects in the world, textual or not, are seen simply as a means of transferring knowledge to humans. While the process is important, the goal is for knowledge to be gained. Sometimes this is the goal of the informer. Other times, there is no informer, and the recipient has the goal of gaining knowledge. Regardless, the goal is to end up with certain knowledge in a human mind, and the process is just another tool.

Information as Knowledge.

The information-as-knowledge viewpoint centers on information as it exists in human minds. In this viewpoint, information-as-thing is simply knowledge in the world that has not yet been transferred to a human. Information-as-process is not information itself, but the noise introduced to information by the act of communication between humans (as described by Reddy, 1979) or communication between one human and the world.

The only way humans can get knowledge is through their senses. Each sense is a separate communication channel that can carry various types of information. Sight conveys shape, color, and brightness information. Hearing conveys frequency and volume information. Rhythm information can come from hearing, sight, or touch. Each of these information channels can introduce noise. Humans use multiple senses combined with previous knowledge to cancel out as much noise as possible, but there is no guarantee that any two humans sense the outside world in the same way.

Ryan's Theory of Conservation of Intelligence

Information (knowledge) cannot be created, it must be

• given by a trusted entity
• generalized from data sets / experience
• tested in the world

1/27/00 class thoughts

Official Lecture Notes

My essay (above) wasn't incredibly coherent, but it does coincide fairly well with Jacob's theory of information. Here's my quick summary of the class discussion:

• Data = ``raw'' facts, as in sensory input
• Information = data that has been processed in terms of existing knowledge, potentially useful for ``informing'' behavior
• Knowledge = stored information
• Wisdom = accrued knowledge, indexed so well that the person does not have to ``think'' to make use of it, and perhaps isn't even aware of its use...expertise in day-to-day living
• One person's information becomes another person's data (in the process of transferring it).

From this I gather that, roughly, information is that which is stored in short-term memory. Of course, and ``information systems'' ``information specialists'' can't place information directly into brains, so they simply process data as much as possible before giving it to a client/user.

There's a strange two-level process at work here that often gets overlooked. As I stated in my essay, all of our data comes in the form of frequencies, amplitudes, temperatures, etc. The perceptive parts of our brains process this data to provide information in the form of words, shapes, etc. to the higher-level parts of our brains. These higher-level structures process the words and shapes into the facts and concepts that we usually consider information. The reverse occurs when we're providing information.

1/31/00 group work

We were given a paperclip, ``salute'', and ``faith''. Our assignment was to determine how much information was in each.

My initial reaction was to measure each on the basis of how much existing knowledge each brought into short-term memory. In this scheme, the paperclip holds very little information, and ``faith'' holds a large amount.

After attempting to clarify the assignment, we decided to rate the items based on the amount of information they could add to a (relatively) blank knowledge base (eg. a baby's mind). In this scheme, the evaluation is reversed. A paperclip is an object that can provide much information by simply looking at it, while ``faith'' is relatively information-free.

We also considered the possibility that (using a completely blank knowledge base) the items all contain the same amount of information: none.

1/31/00 abstract
Lakoff, G., and Johnson, M. (1980). Metaphorical systematicity: highlighting and hiding. In Metaphors we live by (p. 10-13). Chicago: University of Chicago Press.

Summarizes Michael Reddy's conduit metaphor. Emphasizes that the information content of a sentence depends on the reader's bias.

2/1/00 abstract
Machlup, F., and Mansfield, U. (1983). Cultural diversity in studies of information. In F. Machlup and U. Mansfield (Eds.), The study of information: Interdisciplinary messages (p. 3-39). New York: John Wiley.

Compares and contrasts disciplines that study information.