Contextextractor: uma ferramenta de apoio para extração de informações de contexto de artigos de engenharia de software experimental

Abstract

As the number of scientific publications in Empirical Software Engineering grows, so does the need for automated support to locate, organize and summarize information for systematic reviews in the field. Although text mining and information extraction techniques are already being successfully used for secondary data analysis in areas like Biology and Medicine, researchers in software engineering still extract information manually from their papers. This work presents an approach for automated extraction of information for secondary studies in Empirical Software Engineering (ESE). The approach is based on a template that characterizes experimental studies in ESE and uses regular expression rules to obtain the template information fields directly from ESE scientific papers. A tool was developed to implement the approach and two studies were executed to compare its performance against the manual extraction of information from the papers.