Datalab

Modeling Superimposed Information

SPARCE Home ] SI Overview ] Details ] Applications ] [ Modeling ] Development ] Publications ]

 

Dept. of Computer Science
Portland State University

 

We assume you are familiar with terms such as superimposed information, mark, and context.

'To abstract or not' is the question

When modeling superimposed information one is confronted with a simple question: to abstract or not.

Typically, conceptual models hide many details about the information being modeled, but the use of marks makes the two layers of information (superimposed layer and base layer) explicit. Clearly, there are benefits from abstracting away the layers. For example, it hides the fact that base information is potentially heterogeneous and distributed. A consequence of hiding the layers is that the utility of the conceptual model is reduced in later stages of the information life cycle. For example, a logical schema generated from the over-abstracted conceptual schema will be incomplete with respect to marks. Currently, we are working on answering this question about abstraction, and more fundamental questions such as what it means to use a mark and how a mark might be represented in a model.

We are currently working on representing superimposed information using the ER model. Specifically, we are looking at means to represent marks associated with entities, relationships, and attributes of entities and relationships. Because the ER model is not expressive enough for our purpose, we have applied the notion of relationship patterns to enrich the ER model (so we can model the use of marks).

Using Relationship Patterns to Model Superimposed Information

A Framework for Relationship Pattern Languages (describes relationship patterns)

 

In the past, Shawn Bowers has developed the notion of Superimposed Schematics to incorporate marks in an ER-like model. He also developed an application called the Schematics Browser to demonstrate the model.

 

Author: Sudarshan Murthy

Page modified: 13 Sep 2006 10:05 AM