Modeling Associations through Intensional Attributes

Abstract

Attributes, a.k.a. slots or properties, are the main mechanism used to define associations between concepts or individuals modeling real world entities in a knowledge base. Traditionally, an attribute is defined by an explicit statement that specifies the name of the attribute and the entities it associates. This has three main limitations: (i) it is not easy to apply to large amounts of data, even if they share the same characteristics, since explicit definitions are needed for each concept or individual; (ii) it cannot handle future data, i.e., when new concepts or individuals are inserted in the knowledge base their attributes need to be explicitly defined; and (iii) it assumes that the data engineer, or the user that is introducing a new attribute, has access and privileges to modify the respective objects. The above may not be practical in many real ontology application scenarios. We are introducing a new form of attribute in which the domain and range are not specified explicitly but intensionally, through a query that defines the set of concepts or individuals being associated. We provide the formal semantics of this new form of attribute, describe how to overcome syntax constraints that prevent the use of the proposed attribute, study its behavior, show efficient ways of implementation, and experiment with alternative evaluation strategies.