Text summarization techniques applied to source code summary generation and evaluation

Abstract. This dissertation explores the use of text summarization technology for generating automatic descriptions of source code artifacts. Firstly, the thesis reports the results of empirical studies aimed at investigating how developers summarize code artifacts when understanding them. These results are useful in explaining how developers abstract source code, in gaining insights about the comprehension process, in explaining the role of code elements in such a process, and for the development of automatic source code summarization tools. Secondly, the dissertation describes several approaches for creating short and accurate textual descriptions for various types of code entities. The results of the evaluations indicate that text summarization techniques are suitable for automatic source code summarization since developers generally agree with the summaries produced by the proposed methods. Thus, these summaries can be useful for improving software comprehension processes, which usually occur during software maintenance tasks.

• Autores: Aponte Melo, Jairo Hernán
• Tipo de recurso: Doctoral thesis
• Fecha de publicación: 2012
• Institución: Universidad Nacional de Colombia
• Repositorio: Universidad Nacional de Colombia
• Idioma: spa
• OAI Identifier: oai:repositorio.unal.edu.co:unal/20179
• Acceso en línea: https://repositorio.unal.edu.co/handle/unal/20179; http://bdigital.unal.edu.co/10626/
• Palabra clave: 0 Generalidades / Computer science, information and general works; 62 Ingeniería y operaciones afines / Engineering; Software maintenance; Source code summarization; Program comprehension; Automatic software documentation; Empirical research; Mantenimiento de software; Resumen de código fuente; Comprensión de programas; Documentación automática de software; Investigación empírica
• Rights: openAccess
• License: Atribución-NoComercial 4.0 Internacional