Automated support for framework documentation

点击翻译按钮获取中文摘要

Software development organizations rely on application frameworks to build applications. Frameworks provide implemented and tested solutions to recurring requirements, and allow organizations to invest their efforts in innovative, value-adding development through the reuse of software infrastructure. Large frameworks must be supported by extensive technical documentation, which can include reference manuals, tutorials, and code examples. Despite the fact that developing and maintaining framework documentation now requires a major investment, very little is known about the impact of the documentation system on the framework's learnability, and very little automation exists to help organizations manage framework documentation. The goal of this research program is to investigate how automatic techniques can be used to support the development and maintenance of framework documentation. In the short term, we target the development of automated techniques to a) analyze existing documentation and usage data to recommend improvements and updates; b) dynamically adapt documentation systems to users needs based on user modeling techniques; c) automatically create tutorial examples based on software version history mining. For the longer term, we plan to build on our understanding of framework documentation to expand the research to address other pertinent issues in framework usability, and in particular the problem of efficiently querying framework documentation. The research methods for all three objectives will involve a combination of exploratory empirical studies, algorithm and prototype tool development, and validation studies using both active participants and archived project artifacts. The proposed work has the potential to significantly impact how organizations produce and utilize framework documentation and may help organizations shift from a handcrafting approach to a semi-automated documentation creation and management approach supported by advanced software analyses. The research will also involve extensive re-thinking of existing software artifact analysis techniques to account for the unstructured and pedagogical aspects of software documentation.