Improving the Search for Architecture Knowledge in Online Developer Communities

Abstract

When architecting a software system, software engineers search for architectural solutions (e.g. technologies), which fulfill certain requirements. Current approaches for architecture knowledge repositories facilitate learning about different architectural solutions. Nevertheless, the rapid and continuous increase of solution alternatives makes it challenging to manually acquire architecture knowledge and to ensure that this knowledge is up to date. Our goal in this paper is to improve how software engineers search for architecturally relevant information in online developer communities. We developed a new search approach for architecturally relevant information using Stack Overflow as an example of an online developer community. Our search approach differs from a conventional keyword-based search in that it considers semantic information of architecturally relevant concepts in Stack Overflow. We implemented the search approach as a web-based search engine. To show the effectiveness of the search approach compared to a conventional keyword-based search, we conducted an experiment with 16 practitioners. To ensure realism of the experiment, tasks given to practitioners are based on real scenarios identified in a separate interview study. The experiment showed that the new search approach significantly outperforms a conventional keyword-based search. The full paper of this extended Abstract has been published in [So18].