Extremo

Model-driven engineering (MDE) often relies on domain-specific languages (DSLs) to develop complex systems. DSLs usually serve to build models that contain primitives and concepts specifically tailored to a particular domain. The abstract syntax of DSLs is described through a meta-model (which is itself a model), and hence the construction of models and meta-models is a common activity in MDE.

Constructing DSLs requires deep domain understanding, typically involving two roles: a domain expert, who has in-depth knowledge of a particular domain, and a meta-modelling expert, who is experienced in object-oriented design, class-based modelling, and language engineering. However, the latter may lack domain expertise, leading to uninformed decisions and compromised language quality.

Under the hypothesis that reusing heterogeneous knowledge helps build more accurate models more efficiently, this work proposes building a (meta-)modelling assistant. This way, the proposed system represents heterogeneous information sources with a common data model, supports its uniform querying and reusing information chunks for building (meta-)models. Moreover, to understand how and whether modelling assistants help in designing a new DSL, an analytical and empirical evaluation has been conducted. The evaluation tested whether a modelling assistant for information reuse would be useful to obtain more complete and correct (meta-)models, built in a more efficient way, facilitating the labour of the (meta-)modelling expert and demonstrating the benefits of introducing reusability practices in the early phases of software development.

The ideas proposed in this work have been implemented in a prototype application called Extremo, a modelling assistant that helps reuse information from heterogeneous artefacts, like models, meta-models, ontologies, or XML documents.

The tool supports models represented in Ecore/XMI, XSD/XML and ontologies. However new model representations can be supported with the definition of an extension point. To extend the tool follow the next instructions:

Eclipse > New > Other > Plugin project > ... > Finish

In the MANIFEST insert a Dependency to uam.extremo.core.extensions and to uam.extremo.core. Both dependencies give access to the core classes of Extremo.

Add an extension to extremo.core.extensions.assistant and then a new Assistant. Give the assistant a name, a class implementation, and list the file extensions the assistant will be applied to.

A new all-methods-empty class should be created implementing the IFormatAssistant class. This class needs to parse your format to our class structure (shown in the metamodel above).

Adding the dependencies

Selecting the extension

Assistant definition

Class implementation

The tool supports predicate-based searches and custom searches to query the common data model. New model queries can be supported with the definition of an extension point. To extend the tool follow the next instructions:

Eclipse > New > Other > Plugin project > ... > Finish

In the MANIFEST insert a Dependency to uam.extremo.core.extensions and to uam.extremo.core. Both dependencies give access to the core classes of Extremo.

Add an extension to extremo.core.extensions.search and then a new Predicate Based Search / Custom Search. Give the query a name, a class implementation, and a type from the data model to indicate in which level the query will be applied to.

A new all-methods-empty class should be created implementing the ExtensiblePredicateBasedSearch/ExtensibleCustomSearch class. In the predicate based search case, this class needs to return a boolean if the parametrized element matches with a condition. The Custom Search is open to interpretation.