MERA: Meta Language for Software Engineering

A unique feature of MERA is that it can be used to define many different graphical languages by creating a meta-graph in MERA that describes the formalism of the specific graphi-cal language. Therefore, for a specific purpose, a specific MERA is generated. This approach is superior than having a super set of multiple languages which increases chance of the user being overwhelmed by the complexity of the language definition itself. Like the CAKE (Rich, 1991) representation language, the MERA language includes many features that are more commonly found in artificial intelligence knowledge representation languages. However, whereas CAKE applies AI techniques of truth maintenance, many-sorted logic, and frames in a multi-layered architecture to software engineering , MERA applies semantic networks, sets, ontological markers for intensionality, and user-adaptation to visual representation of both software engineering information as well as AI natural language processing data such as syntax and semantics. Following the entity-relationship data model (Chen, 1976), which presents a unified view of database information, the basic building blocks of MERA are entities, relations, and attributes. Entities are represented as an icon and relations as a line (solid or dotted). Attributes are name-value pairs and are associated with objects , which are defined to be all entities and relations. Attributes have type constraints, range constraints on numeric types, a constraint on when values are required, a set constraint that specifies whether an attribute can consist of a set of values, and an optional default value. Relations have a source and a sink which are objects, so a relation can be used to relate two other relations (or even itself). Objects (entities and relations) have names and constraints on when a name is required. Relations also have constraints on the types of their sources and sinks, connectiv-ity (e.g. 1-m means that any object can serve as the source of only one such relation, but many such relations and sink into one object), reflexivity (whether an object can serve as both the source and sink of the same relation), and multiple (whether more than one such relation can connect the same two source and sink objects). An example of a MERA graph is shown in Figure 1. This MERA is a versatile graphical language which can be used for modeling and analysis of the user requirements, for building a prototype, for representing designer's knowledge , for controlling design process, etc. The current version of MERA …