| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 820.33 KB | Adobe PDF |
Authors
Abstract(s)
A presente dissertação resulta da necessidade de classificação de um conjunto de requisitos de
software crítico não categorizados, em categorias distintas através de métodos tradicionais e
de ferramentas de Natural Language Processing, no contexto do projeto PROVA - Platform for
Software Verification and Validation, levado a cabo pela empresa EDUCED.
Justifica-se este projeto com a necessidade de existir um pré-processamento do texto referente
aos requisitos de software crítico que a plataforma do PROVA irá suportar, mas também devido
ao facto de a classificação dos requisitos de software ser um trabalho manual, feito por um
engenheiro de requisitos podendo levar bastante tempo a ser concluído.
Nesta dissertação é apresentado, detalhadamente, um método para a criação de um classificador de requisitos de software crítico baseado em métodos de processamento de texto,
complementados com ferramentas de processamento de linguagem natural para uma classificação mais precisa. O sistema irá permitir a entrada dos requisitos de software crítico em vários
formatos e retornará no final a classificação individual de cada requisito.
A classificação individual do requisito é obtida através do recurso a um parser que recebe o
texto do requisito de software e procura palavras-chave e/ou expressões que auxiliem na identificação de cadeias de símbolos, que de acordo com a gramática definida e através de métodos de comparação dessas cadeias irá concluir em qual categoria o requisito pode ser classificado. De referir ainda que quando o texto do requisito de software chega ao parser, foi anteriormente complementado com o etiquetas sintáticas criadas para uma melhor classificação do requisito de software.
The present dissertation results from the need of classifying a set of non categorized critical software requirements in different categories using traditional methods and Natural Language Processing tools, under the PROVA project - Platform for Software Verification and Validation, carried out by EDUCED company. This project is justified by the need of a preprocessing of the text referring to the critical software requirements PROVA platform will support, but also due to the fact that the classification of software requirements is a manual task performed by a requirements engineer that can take a long time to complete. In this dissertation will be presented in detail a method to create a classifier of critical software requirements with software-based methods of text processing, supplemented with natural language processing tools for a more accurate classification. The system will allow the entry of critical software requirements in various formats and will return at the end, the individual classification of each software requirement. The individual classification of the software requirement is obtained by using a parser that receives the software requirement text and searches for keywords and/or sentences that assist in the identification of strings of symbols, which according to the defined grammar and using string comparison methods will conclude in which category the software requirement fit. Also note that when the text reaches the software requirement parser, it was previously complemented with syntactic labels designed for better software requirement classification.
The present dissertation results from the need of classifying a set of non categorized critical software requirements in different categories using traditional methods and Natural Language Processing tools, under the PROVA project - Platform for Software Verification and Validation, carried out by EDUCED company. This project is justified by the need of a preprocessing of the text referring to the critical software requirements PROVA platform will support, but also due to the fact that the classification of software requirements is a manual task performed by a requirements engineer that can take a long time to complete. In this dissertation will be presented in detail a method to create a classifier of critical software requirements with software-based methods of text processing, supplemented with natural language processing tools for a more accurate classification. The system will allow the entry of critical software requirements in various formats and will return at the end, the individual classification of each software requirement. The individual classification of the software requirement is obtained by using a parser that receives the software requirement text and searches for keywords and/or sentences that assist in the identification of strings of symbols, which according to the defined grammar and using string comparison methods will conclude in which category the software requirement fit. Also note that when the text reaches the software requirement parser, it was previously complemented with syntactic labels designed for better software requirement classification.
Description
Keywords
Classificador Etiquetas Sintáticas Gramática Parser Processamento de Linguagem Natural Requisitos de Software