The Formal Semantics of the Domain Specific Modeling Language for Multiagent Systems

Abstract

Abstract. Recently, associated with the increasing acceptance of agentbased computing as a novel computing paradigm a lot of research has been addressed to develop mechanisms and methods to support the agent-based development of complex software systems. Especially the idea to define agent-oriented languages on a more abstract level through metamodels is recently often applied. However, the metamodel's opportunity to express the language's semantics are restricted as only concepts and their relationships to each other can be defined within the metamodel. This paper discusses an approach to formalize the semantics of Dsml4mas-a modeling language for multiagent systems-to support the system designer in validating and verifying the generated design.

Interoperability of MAS DSMLs via horizontal model transformations

“…Due to the difficulty encountered on repeating those vertical model transformation steps, current MAS DSML proposals (e.g. [13][14][15], [17], [19]) mostly support the execution of modeled agents on just one agent platform. Rarely, two different platforms are supported (e.g.…”

“…DSML4MAS [5], [13] is perhaps one of the first complete MAS DSMLs in which a PIMM, called PIM4Agents, provides an abstract syntax for different aspects of agent systems. Similar to SEA_ML's viewpoints, both internal behavior model of agents and agent interactions in a MAS are covered by PIM4Agents views / aspects.…”

“…specific modeling languages (DSMLs) (e.g. [12][13][14][15][16][17][18][19][20]) for facilitating the development of MASs. DSLs / DSMLs [21][22][23] have notations and constructs tailored toward a particular application domain (e.g.…”

“…Proposed MAS DSMLs such as [13], [17], [19] usually support modelling both the static and the dynamic aspects of agent software from different MAS viewpoints including agent internal behaviour model, interaction with other agents, use of other environment entities, etc. Within this context, abstract syntaxes of the languages are represented with metamodels covering those aspects and required viewpoints to some extent.…”

“…Differentiating from the existing MAS DSML studies (e.g. [13], [16], [17], [19], [20]), our proposal is based on determining entity mappings and building horizontal M2M transformations between the metamodels of MAS DSMLs which are in the same abstraction level. In this paper, we also investigate the applicability of the proposed DSML interoperability approach by constructing horizontal transformations between two full-fledged agent DSMLs called SEA_ML [19] and DSML4MAS [5] respectively.…”