Método de conversión de un diálogo controlado a un discurso en UN-Lencep
Cargando...
Autores
Arévalo Camacho, Wiliam Alfonso
Director
Tipo de contenido
Document language:
Español
Fecha
Título de la revista
ISSN de la revista
Título del volumen
Documentos PDF
Resumen
Resumen: El análisis de requisitos es una etapa del ciclo de vida del software que comprende las tareas de educción, modelado, validación y especificación de los requisitos. Adicionalmente, la especificación de requisitos debe ser correcta, no ambigua, completa, consistente, jerarquizada, verificable, modificable y trazable. Para realizar este proceso se requiere un contacto permanente con el interesado, de forma que toda la información se constate con él. Existen diversas técnicas para obtener la información necesaria para una especificación de requisitos, pero la que más se utiliza es la entrevista, la cual es un diálogo entre el analista y el interesado. Sin embargo, en un diálogo es común encontrar factores que afectan el entendimiento entre los actores. Esto hace que la tarea de obtener la información necesaria del interesado sea una de las más complejas en el proceso de especificación de requisitos. En procura de disminuir la ambigüedad existente en el diálogo, se desarrollaron los lenguajes controlados, que son subconjuntos del lenguaje natural. Los lenguajes controlados poseen una estructura similar al lenguaje natural, con reglas léxicas, reglas gramaticales, signos y palabras. Los lenguajes controlados tienen diversos usos. En la especificación de requisitos de software se encuentra el lenguaje controlado UN-Lencep, que permite presentar el discurso del interesado de una forma que se pueda validar. Adicionalmente, presenta la información de manera concreta, inambigua y completa. Sin embargo, la información se debe obtener mediante el diálogo con el interesado, dejando en manos del analista la identificación de los elementos necesarios para la estructuración del discurso, lo que posibilita la aparición de errores. Para reducir esta problemática, se propone, en esta Tesis, la estructuración de una secuencia ordenada de preguntas y la definición de las reglas necesarias para convertir las respuestas en el discurso del interesado, expresado en UN-Lencep. Adicionalmente, en un prototipo funcional se incluyen estos elementos y se valida con la especificación de algunos proyectos que requieren el desarrollo de una aplicación de software.
Abstract: The requirements analysis is a phase of software development life-cycle. It comprises: requirements elicitation, modeling, validation and specification. Also, the requirements specification should be: correct, unambiguous, complete, consistent, hierarchical, verifiable, modifiable, and traceable. This process requires a permanent contact with the stakeholder, in such way that he can verify the information. There are several techniques for obtain information for the requirements specification, but the most used is the interview, which is a dialog between the analyst and the stakeholder. In such dialog the stakeholder has the information about the software domain and the analyst knows how make the software specification. However, some problems arise in dialogs that affect the understanding between their actors. They make most complex the task of requirements specification process, in order to obtain the needed information from the stakeholder. The controlled languages (subsets of natural language) help to decrease the ambiguity in the dialog. The controlled languages have a similar structure to the natural language, with lexical and grammatical rules, signs, and words. The controlled languages have several uses. For the software requirements specification, a controlled language called UN-Lencep is used, for allowing elaboration and validation of the stakeholder discourse. UN-Lencep presents the information in a concrete, unambiguous, and complete way. However, the information must be obtained by using a dialog with the stakeholder, but the analyst must identify the elements of the discourse structure, and this is a possible cause of mistakes in the process. As a way to reduce this problem, we propose, in this Thesis, the organization of an ordered sequence of questions that allow guidance to analyst on the identification of the needed elements for creating the discourse. Also, we establish a set of rules for translating the answers into a UN-Lencep stakeholder discourse. We include these features into a functional prototype, which we use in a lot of projects that needs to develop a software application, so we validate the functionality of all elements
Abstract: The requirements analysis is a phase of software development life-cycle. It comprises: requirements elicitation, modeling, validation and specification. Also, the requirements specification should be: correct, unambiguous, complete, consistent, hierarchical, verifiable, modifiable, and traceable. This process requires a permanent contact with the stakeholder, in such way that he can verify the information. There are several techniques for obtain information for the requirements specification, but the most used is the interview, which is a dialog between the analyst and the stakeholder. In such dialog the stakeholder has the information about the software domain and the analyst knows how make the software specification. However, some problems arise in dialogs that affect the understanding between their actors. They make most complex the task of requirements specification process, in order to obtain the needed information from the stakeholder. The controlled languages (subsets of natural language) help to decrease the ambiguity in the dialog. The controlled languages have a similar structure to the natural language, with lexical and grammatical rules, signs, and words. The controlled languages have several uses. For the software requirements specification, a controlled language called UN-Lencep is used, for allowing elaboration and validation of the stakeholder discourse. UN-Lencep presents the information in a concrete, unambiguous, and complete way. However, the information must be obtained by using a dialog with the stakeholder, but the analyst must identify the elements of the discourse structure, and this is a possible cause of mistakes in the process. As a way to reduce this problem, we propose, in this Thesis, the organization of an ordered sequence of questions that allow guidance to analyst on the identification of the needed elements for creating the discourse. Also, we establish a set of rules for translating the answers into a UN-Lencep stakeholder discourse. We include these features into a functional prototype, which we use in a lot of projects that needs to develop a software application, so we validate the functionality of all elements