TOSCANA management system for conceptual data

Abstract

TOSCANA is a software for the visualization of data with nested line diagrams. It is a tool for navigating in databases and for the retrieval of objects in a database. It enables the user to discover relationships and implications between attributes.TOSCANA is based on Formal Concept Analysis as introduced in [Wi82]. This method formalizes a relation I between a set G of objects and a set M of attributes by the formal context (G, M, I). A formal context can be handled as a table or a view of a relational database. The formal context determines its concept lattice which is the structure of the relation subconcept-superconcept.The concept lattices are graphically represented as line diagrams. For the basic notions of Formal Concept Analysis see also [DP90, chapter 11] The line diagrams for practical applications get very large. TOSCANA uses nested line diagrams which were introduced in [Wi84] (see also [Wi89]). We split a given context into parts, draw a line diagram of each part and then nest these line diagrams into each other. This gives a simplified diagram because groups of parallel lines are replaced by one line and the ellipses around the small diagrams (see Figure 2). TOSCANA automatizes the nesting of the diagrams and allows a flexible choice of the diagrams. As mentioned above, the formal context can be kept as a table or view in a database. TOSCANA selects the information for the nested line diagram from such a database. Thus, TOSCANA helps to analyze large sets of data and enables the user to navigate through this database.The ideas for TOSCANA are first described in [VWW91]. The further development of these ideas is reflected in [SSVWW93], [KSVW94], [VW95]. This last paper was presented to the Graph Drawing conference in 1994 at Princeton.We will demonstrate one basic direction of applying TOSCANA. The example (see 1, 2, 3) shows how TOSCANA is used to structure a set of documents. The documents, which take the role of the objects of the formM context, are related to search keys, which are the attributes. The documents are German law documents and technical norms. The search keys are grouped into themes. Each theme is represented by a separate line diagram. Now a users sees the documents not only for one search key but for the whole group of keys for the selected theme. The documents were unfolded from the very specific documents to the general ones. The combination of themes Mlow complex selections and searches.The screen dumps in figure 1, 2, 3 show the functionallity of TOSCANA for navigating within the diagrams and selecting objects. The functions to improve