FCA assisted IF Channel Construction towards Formulating Conceptual Data Modeling

In this paper we explore how IF (Barwise-Seligman’s information flow theory) [3] and FCA (Formal Concept Analysis) [20] may be used to help conceptual data modeling. A fundamental observation we make is that the process of conceptual data modeling is a distributed system in the sense of Barwise-Seligman’s information flow theory, i.e., the Channel theory [3], and therefore the process of conceptual data modeling per se, the correctness of it, and thus the usefulness of the resultant database depends solely on the existence of a relevant information channel (hereafter IF channel) that takes the real world domain being modelled and the conceptual data model as components. Based upon this conviction, we explore how such an IF channel may be constructed and how it guides conceptual data modeling. The main findings through this work are as follows • Using IF and FCA for conceptual data modeling, domain-dependent knowledge is still needed in establishing a set of original correspondences between objects in the real world and those in a conceptual data model. This constitutes the basis for further modeling work; • IF and FCA help the modeling in that all those that are required by the existence of an IF channel guide the process of modeling and eventually determine the correctness of it. This includes: o Helping check the correctness of the domain-dependent knowledge reflected in the original set of correspondences; o Guiding the solicitation of knowledge about the domain from the domain expert throughout the process of modeling. We use a simple example to illustrate our points. Key-Words: Conceptual Modeling, Information Flow, Infomorphisms, Concept Lattice, FCA