A domain-specific modeling language for scientific data composition and interoperability

Abstract

Domain-Specific Modeling Languages (DSMLs) can offer assistance to domain experts, who may not be computer scientists, by providing notations and semantic constructs that align with abstractions from a particular domain. In this paper, we describe our design and application of a DSML in the area of data composition and interoperability. In particular, we introduce our recent effort to design a DSML to assist with interoperability issues across scientific software applications (e.g., composing scientific data in different file structures and integrating scientific data with data gathering devices). Currently, several different scientific data file specifications have been proposed (e.g., CID, netCDF, and HDF). Each file specification is optimized to manage a specific data type efficiently. Thus, each file specification has evolved with slightly different notions and implementation technologies. These differences led to the need for an environment that provides interoperability among the different specification formats. In this paper, we introduce our framework, supported by a DSML, that provides functionality to visually model the data composition and integration concepts independent from a particular data file specification.