/IssueAnalysis

SUMMIT

This repository contains the code for the CSCW 2023 paper SUMMIT: Scaffolding Open Source Software Issue Discussion Through Summarization.

Other relevant links: arXiv, CSCW Blog Post.

Code

The code is divided into three branches: extension, backend and engine.

  • extension branch contains the code for the extension user interface which works with the GitHub Issues page for Chromium based browsers.
  • backend branch contains the code for a Python backend which stores the summaries and other metadata in a Postgres database.
  • engine branch contains the code and dataset for training the summarization model and inference service.

engine should be installed first, followed by backend and then extension.

Prototyping

The descriptions for each of the workflows is described and tracked in GitHub issues. The included Framer links need an account to access.

Tool Navigation Workflows

  1. See all sentences of a certain type (Framer, GitHub)

  2. See summarized comments of a subsummary (Framer, GitHub)

  3. See the definition of an information type (Framer, GitHub)

Summary Contribution Workflows

  1. Post a new comment (Framer, GitHub): Also see 10. Top level info type summary gets automatically added below.

  2. Edit the top level summary (Framer, GitHub)

  3. Edit an existing subsummary (Framer, GitHub)

  4. Edit existing highlights (information types) (Framer, GitHub)

  5. Generate a subsummary of a group of unsummarized comments without posting a new comment (Framer, GitHub)

  6. New info type is added to top level summary (Framer, GitHub)

  7. Top level info type summary gets automatically added (Framer, GitHub)

  8. Top level summary has a sentence that comes from comments that are edited/removed (Framer, GitHub)

  9. Edit existing highlights by clicking comment (Framer, GitHub)