The authors note the considerable research and engineering efforts to create numerous software assistants (SAs) that have higher autonomy and intelligence for various stages and aspects of the software life cycle, including software design, development, and management. Their survey identifies 112 studies in 7580 publications based on criteria that spans trends and goals, task performed, modes of interaction with users, technologies used, automation pattern, and embedded knowledge.
The paper has seven sections, 72 references, and an appendix, “Primary Studies,” which lists 112 publications without any commentary. The first section introduces and defines SA (bots) and research directions. The second section redefines SA and its use in the context of software engineering (SE). The third section outlines the research method. The study starts with a set of four questions and states “inclusion and exclusion” criteria that the reported bot must serve. Subsequent sections tabulate selected conferences, journals, and keywords, along with a snowballing criteria and the data extraction method used to select the studied articles.
In Section 4, results from the analysis and classification of bots is presented using a set of tables relating to sources, purpose, tasks performed, types, and data sources, with a focus on information analysis, action implementation, information acquisition, and decision selection. In the next two sections, the authors discuss limitations, the validity of the results, and scope for further work before presenting their conclusions in the last section.
It could be a useful read for researchers interested in bots.