Gustavo Cabral | The University of Tokyo (original) (raw)
Papers by Gustavo Cabral
The use of models in more concrete phases of the software development process is increasing. Most... more The use of models in more concrete phases of the software development process is increasing. Most of current MDE researches concentrate on transformation and conformance checking of Computational Models. However, Computational Independent Models (CIM) and their relations to concrete models is still unclear. This paper explores Requirements Engineering problems in the light of MDE. Application domain, requirement and specification definitions are viewed based on MDE concepts, such as metametamodel, metamodel, transformation and ontologies. From this analysis improvements in MDE are proposed so it can handle requirements concepts, such as softgoals.
The adequacy of MDE technologies to handle system artifacts from early development stages is stil... more The adequacy of MDE technologies to handle system artifacts from early development stages is still an on going research area. Part of the problem is that domain, business and system requirements are usually defined in natural language and therefore context-free grammar parsers cannot be used for model injection (text-to-model). This paper explores the combined use of Natural Language Processing and Model-based Engineering and propose a transformation library that allows querying of grammatical relations over text. Natural Language concepts, such as Partof-Speech and Phrasal Structure, are represented through a metamodel enabling the identification of concepts useful to define reasoning strategies for analyzing textual documents.
Model-driven Engineering brings a viable level of formalism to represent artifacts. It enables mo... more Model-driven Engineering brings a viable level of formalism to represent artifacts. It enables model visualization, transformation and property checking. However, the adequacy of MDE to handle system artifacts from early development stages is problematic since these are usually defined in natural language and therefore context-free grammar parsers cannot be used to accomplish model injection (text-to-model). This paper proposes a strategy that combines Natural Language Processing and MDE to enable the extraction of a Knowledge model from the input NL text. The Knowledge model is an Ontology with classes, relations and instances that is populated with the input text. We define a supervised approach that requires the selection of the correct translation of sentences among suggestions built by a tool. We apply the tool to sentences from a specification document and create a Domain Knowledge model, which also assists us to understand new concepts from the analyzed field.
Automatic generation of formal specifications from requirement reduces cost and complexity of for... more Automatic generation of formal specifications from requirement reduces cost and complexity of formal models creation. Thus, the generated formal model brings the possibility to carry out system property verification. This paper proposes a Controlled Natural Language (CNL, a subset of English), use case specification templates, and a strategy and tool support to generate process algebraic formal models (in CSP notation) from use cases specified using the templates and CNL. We define templates that represent requirement at different levels of abstraction. Moreover, a refinement notion is defined based on events mapping between abstract and concrete models.
Domain Specific Languages (DSL) play an important role in the Model-driven Engineering (MDE). The... more Domain Specific Languages (DSL) play an important role in the Model-driven Engineering (MDE). Their design can be accomplished by defining metamodel and concrete syntax models. Moreover, transformations can be implemented through transformation languages. However, the process of designing the operational semantics for a DSL should be straightforward. The cost of manipulating and transforming models into programming languages code (GPLs, such as Java) should be minimized, or even nil, since the goal is to use DSLs and not only maintain them. Therefore, we propose a framework called L4J (Language for Java) that allows the definition of DSLs and their behavior, enabling them to be executed. The framework is fully based on the Java language making it a low level of abstraction solution when compared to existing solutions. Metamodel and syntax concepts are defined in Java making use of object-orientation and code annotation through the L4J API. A transformation component that permits the coding and debugging of transformation is provided. The injection and extraction operations are also implemented. We validate the framework by using it in a test case automation case study.
Testing is not enough to achieve a correct and bug free system; it is necessary to maintain a coh... more Testing is not enough to achieve a correct and bug free system; it is necessary to maintain a cohesive set of artifacts, such as use cases, design diagrams, code and test cases. If this elements are loosely coupled, it is very probable that the implementation will not attend the clients expectations. In this context, we provide a model-based DSL editor to assist the engineer in the requirement specification edition and transform it into a formal use model in the CSP notation. This use model is further combined with test directives, or test purposes, to generate test cases using the FDR refinement checker. In addition, a strategy to execute the generated test cases in the Eclipse test platform is proposed.
O objetivo deste trabalho é desenvolver uma ferramenta que vise à automação da aplicação de trans... more O objetivo deste trabalho é desenvolver uma ferramenta que vise à automação da aplicação de transformações sobre modelos[17] definidos na notação UML-RT[16]. Baseado em um conjunto de regras de transformação[5], podemos realizar mudanças em um modelo independente de plataforma (PIM, Plataform Independent Model) agregando informação durante o ciclo de vida de seu projeto. Facilitando assim o processo de definição de sua plataforma (PSM, Plataform Specific Model). A automação da transformação entre estes modelos será possivelmente realizada implementada como um plug-in para a ferramenta Rational Rose RT.
Automatic generation of formal specifications from requirement reduces cost and complexity of for... more Automatic generation of formal specifications from requirement reduces cost and complexity of formal models creation. Thus, the generated formal model brings the possibility of accomplishing system properties validation. This paper proposes a Controlled Natural Language (CNL), use case specification templates, and a strategy and tool support to generate process algebraic formal models (in CSP notation) from use cases specified using the templates and CNL. We define templates that represent requirement at different levels of abstraction. Moreover, a refinement notion is defined based on events mapping between abstract and concrete models.
The automatic generation of formal specifications from requirements suppresses the complexity of ... more The automatic generation of formal specifications from requirements suppresses the complexity of formal models manual creation and reveals the immediate benefits of its usage, such as the possibility to carry out refinements, and property verification, which contributes to project cost reduction and quality improvement. This paper proposes a Controlled Natural Language (CNL), a subset of English, used to write use case specifications according to a template. From these use cases a complete strategy and tools enable the generation of process algebraic formal models in the CSP notation. We define templates that represent requirements at different levels of abstraction, capturing different views of the system behavior. Moreover, a refinement notion is defined to connect the generated CSP models through an event mapping relation between abstract and concrete models. This notion is further applied to detail use case specifications and to automate its execution.
The use of models in more concrete phases of the software development process is increasing. Most... more The use of models in more concrete phases of the software development process is increasing. Most of current MDE researches concentrate on transformation and conformance checking of Computational Models. However, Computational Independent Models (CIM) and their relations to concrete models is still unclear. This paper explores Requirements Engineering problems in the light of MDE. Application domain, requirement and specification definitions are viewed based on MDE concepts, such as metametamodel, metamodel, transformation and ontologies. From this analysis improvements in MDE are proposed so it can handle requirements concepts, such as softgoals.
The adequacy of MDE technologies to handle system artifacts from early development stages is stil... more The adequacy of MDE technologies to handle system artifacts from early development stages is still an on going research area. Part of the problem is that domain, business and system requirements are usually defined in natural language and therefore context-free grammar parsers cannot be used for model injection (text-to-model). This paper explores the combined use of Natural Language Processing and Model-based Engineering and propose a transformation library that allows querying of grammatical relations over text. Natural Language concepts, such as Partof-Speech and Phrasal Structure, are represented through a metamodel enabling the identification of concepts useful to define reasoning strategies for analyzing textual documents.
Model-driven Engineering brings a viable level of formalism to represent artifacts. It enables mo... more Model-driven Engineering brings a viable level of formalism to represent artifacts. It enables model visualization, transformation and property checking. However, the adequacy of MDE to handle system artifacts from early development stages is problematic since these are usually defined in natural language and therefore context-free grammar parsers cannot be used to accomplish model injection (text-to-model). This paper proposes a strategy that combines Natural Language Processing and MDE to enable the extraction of a Knowledge model from the input NL text. The Knowledge model is an Ontology with classes, relations and instances that is populated with the input text. We define a supervised approach that requires the selection of the correct translation of sentences among suggestions built by a tool. We apply the tool to sentences from a specification document and create a Domain Knowledge model, which also assists us to understand new concepts from the analyzed field.
Automatic generation of formal specifications from requirement reduces cost and complexity of for... more Automatic generation of formal specifications from requirement reduces cost and complexity of formal models creation. Thus, the generated formal model brings the possibility to carry out system property verification. This paper proposes a Controlled Natural Language (CNL, a subset of English), use case specification templates, and a strategy and tool support to generate process algebraic formal models (in CSP notation) from use cases specified using the templates and CNL. We define templates that represent requirement at different levels of abstraction. Moreover, a refinement notion is defined based on events mapping between abstract and concrete models.
Domain Specific Languages (DSL) play an important role in the Model-driven Engineering (MDE). The... more Domain Specific Languages (DSL) play an important role in the Model-driven Engineering (MDE). Their design can be accomplished by defining metamodel and concrete syntax models. Moreover, transformations can be implemented through transformation languages. However, the process of designing the operational semantics for a DSL should be straightforward. The cost of manipulating and transforming models into programming languages code (GPLs, such as Java) should be minimized, or even nil, since the goal is to use DSLs and not only maintain them. Therefore, we propose a framework called L4J (Language for Java) that allows the definition of DSLs and their behavior, enabling them to be executed. The framework is fully based on the Java language making it a low level of abstraction solution when compared to existing solutions. Metamodel and syntax concepts are defined in Java making use of object-orientation and code annotation through the L4J API. A transformation component that permits the coding and debugging of transformation is provided. The injection and extraction operations are also implemented. We validate the framework by using it in a test case automation case study.
Testing is not enough to achieve a correct and bug free system; it is necessary to maintain a coh... more Testing is not enough to achieve a correct and bug free system; it is necessary to maintain a cohesive set of artifacts, such as use cases, design diagrams, code and test cases. If this elements are loosely coupled, it is very probable that the implementation will not attend the clients expectations. In this context, we provide a model-based DSL editor to assist the engineer in the requirement specification edition and transform it into a formal use model in the CSP notation. This use model is further combined with test directives, or test purposes, to generate test cases using the FDR refinement checker. In addition, a strategy to execute the generated test cases in the Eclipse test platform is proposed.
O objetivo deste trabalho é desenvolver uma ferramenta que vise à automação da aplicação de trans... more O objetivo deste trabalho é desenvolver uma ferramenta que vise à automação da aplicação de transformações sobre modelos[17] definidos na notação UML-RT[16]. Baseado em um conjunto de regras de transformação[5], podemos realizar mudanças em um modelo independente de plataforma (PIM, Plataform Independent Model) agregando informação durante o ciclo de vida de seu projeto. Facilitando assim o processo de definição de sua plataforma (PSM, Plataform Specific Model). A automação da transformação entre estes modelos será possivelmente realizada implementada como um plug-in para a ferramenta Rational Rose RT.
Automatic generation of formal specifications from requirement reduces cost and complexity of for... more Automatic generation of formal specifications from requirement reduces cost and complexity of formal models creation. Thus, the generated formal model brings the possibility of accomplishing system properties validation. This paper proposes a Controlled Natural Language (CNL), use case specification templates, and a strategy and tool support to generate process algebraic formal models (in CSP notation) from use cases specified using the templates and CNL. We define templates that represent requirement at different levels of abstraction. Moreover, a refinement notion is defined based on events mapping between abstract and concrete models.
The automatic generation of formal specifications from requirements suppresses the complexity of ... more The automatic generation of formal specifications from requirements suppresses the complexity of formal models manual creation and reveals the immediate benefits of its usage, such as the possibility to carry out refinements, and property verification, which contributes to project cost reduction and quality improvement. This paper proposes a Controlled Natural Language (CNL), a subset of English, used to write use case specifications according to a template. From these use cases a complete strategy and tools enable the generation of process algebraic formal models in the CSP notation. We define templates that represent requirements at different levels of abstraction, capturing different views of the system behavior. Moreover, a refinement notion is defined to connect the generated CSP models through an event mapping relation between abstract and concrete models. This notion is further applied to detail use case specifications and to automate its execution.