Characterizing Architecture Related Posts and Their Usefulness in Stack Overflow

TL;DR

This study analyzes architecture-related posts on Stack Overflow to categorize them, assess their usefulness, and provide guidelines for better question formulation to improve response rates.

Contribution

It offers a detailed classification of architecture-related questions on SO and insights into factors influencing their usefulness and response rates.

Findings

Nine core categories of architecture questions identified

Clear descriptions and diagrams increase answer count

Poorly structured questions tend to receive fewer answers

Abstract

Context: Stack Overflow (SO) has won the intention from software engineers (e.g., architects) to learn, practice, and utilize development knowledge, such as Architectural Knowledge (AK). But little is known about AK communicated in SO, which is a type of high-level but important knowledge in development. Objective: This study aims to investigate the AK in SO posts in terms of their categories and characteristics as well as their usefulness from the point of view of SO users. Method: We conducted an exploratory study by qualitatively analyzing a statistically representative sample of 968 Architecture Related Posts (ARPs) from SO. Results: The main findings are: (1) architecture related questions can be classified into 9 core categories, in which "architecture configuration" is the most common category, followed by the "architecture decision" category, and (2) architecture related questions that provide clear descriptions together with architectural diagrams increase their likelihood of getting more than one answer, while poorly structured architecture questions tend to only get one answer. Conclusions: Our findings suggest that future research can focus on enabling automated approaches and tools that could facilitate the search and (re)use of AK in SO. SO users can refer to our proposed guidelines to compose architecture related questions with the likelihood of getting more responses in SO.