Back
2025INF004
Doctoral thesis

Mapping and reifying the software documentation landscape

  • 2025
Software documentation landscape
Mining software documentation
Mapping documentation sources
Modern software documentation
Discord
Uml
English Every software system (ideally) comes with one or more forms of documentation, a fundamental asset of every software project. Documentation is deemed useful for different stages of the software lifecycle (e.g., design, implementation, maintenance) and is created and maintained by different stakeholders for different target audiences. In all this heterogeneity, multiple platforms (e.g., instant messaging, Q&A websites, social networks) host and contribute a substantial amount of documentation. Developers resort to Discord, StackOverflow, and a mix of Google, emails, and Tweets to "get things done". We refer to this evolving heterogeneous collection of information sources as the documentation landscape of a software system. A systematic approach to discover, classify, and integrate these sources in a unifying framework is fundamental to understand the paradigm shifts and complex phenomena affecting modern software documentation. To look at software documentation from a new perspective, we propose an approach to map and reify the documentation landscape of a software system. We identify and retrieve the sources of documentation used by developers and investigate some of the many forms that documentation sources can take. We provide a holistic interpretation of the software documentation landscape and the phenomena determining its evolution. Traditional documentation is becoming a thing of the past, drifting towards fast (but also extremely volatile and noisy) communication channels, hosts of valuable chunks of the modern software documentation landscape. We complement our approach with an evolutionary analysis that explores in detail software documentation in instant messaging applications and UML artifacts. Through a mix of object oriented domain modeling and software visualization techniques, we characterize documentation sources according to form and content, analyzing their interplay. Finally, we discuss the current limitations and the potential future of the software documentation landscape, highlighting the need of a machine-readable specification to fully automate our mapping approach.
Collections
Language
  • English
Classification
Computer science and technology
License
License undefined
Open access status
green
Identifiers
Persistent URL
https://n2t.net/ark:/12658/srd1331542
Statistics
Document views: 12 File downloads:
  • 2025INF004.pdf: 32