Tuesday, September 16, 2003

The Intertext of Entity/Relationship Diagrams

After reviewing the vade mecum of entity relationship diagrams and relational models (Date, 1995), I have a few questions and concerns. We currently ask a lot from our databases and commentary has been written on the importance of applying the human information behaviour literature to information retrieval technology (Hewins, 1990). The question I have is whether or not the technology can deliver.

My first concern comes from the nature of databases themselves and our efforts to make them more efficient. The process of normalization strips away redundant information in a manner that is both reversible and lossless. I can’t help but think that much of the information that is removed may not be redundant at all. By changing the organization of our “tupples” the user may become mired in a sea of unfamiliar terms. Imagine two different fields in the same database: one for the English word and one for the corresponding French word. By relegating the French word to a lookup table or thesaurus it would seem that the overall efficiency of the system is compromised… at least for the French.

An additional concern is the nature of the semantic information contained within a database. Ideally each field in a database contains a scalar—a single value with a static meaning. While this situation is fine for persistent financial data, static meanings are considerably more difficult to attain in IR systems. Early semioticians realized that the links between specific words and their meanings are arbitrary and determined primarily through language (Burr, 1995; Wittgenstein, 1958). Databases however lock in a static meaning. Indeed, in particular instances (like unique keys, etc.) databases demand a constant semantic significance. This mismatch between the demands of computer science and the constraints of human cognition may lead user queries astray. Instead of producing results that are relevant, the query falls through the gaps of intertextuality inherent in the database design.

Perhaps my most grave concern is for our handmaidens of database design: entity/relationship diagrams. Date (1995) notes that in E/R diagrams, relationships cannot be entities. This statement seems to suggest that relationships also lack the features of entities such as properties. In an E/R diagram, a relationship can exist in a state of sub- or superiority but we can never know the intimate details. It seems that many of our current understandings of information as a type of process to affect cognitive change (Belkin, 1978; Buckland, 1991; Dervin, 1983) depends on exactly these sort of details. In the E/R paradigm of sanitized relationships there is no room for potentially vital information. Sure we know who an individual’s boss is, but can we discern details about their relationship? Who buys the morning coffee? What tasks does the boss _really_ delegate? What other tacit details crucial to understanding something get left on the cutting room floor when we concoct our E/R diagrams?


Belkin, N. J. (1978). Information concepts for information science. Journal of Documentation, 34(1), 55-85.
Buckland, M. K. (1991). Information as thing. Journal of the American Society for Information Science, 42(5), 351-360.
Burr, V. (1995). An introduction to social constructionism. New York: Routledge.
Date, C. J. (1995). An introduction to database systems (6th ed. ed.). Reading Mass.: Addison-Welsley.
Dervin, B. (1983). Information as a user construct: The relevance of perceived information needs to synthesis and interpretation. In S. A. Ward & L. J. Reed (Eds.), Knowledge structure and use: Implications for synthesis and interpretation (pp. 155-183). Philadelphia: Temple University Press.
Hewins, E. T. (1990). Information need and use studies. Annual review of information science and technology, 25, 145-172.
Wittgenstein, L. (1958). Philosophical investigations (G. E. M. Anscombe, Trans.). New York: Macmillan.


Post a Comment

Subscribe to Post Comments [Atom]

<< Home