Data and results for the paper: "From Bugs to Benefits: Leveraging Crowd-Sourced Issue Data to Improve User Stories with CrUISE-AC"

This page provided the following files used in the study "From Bugs to Benefits: Leveraging Crowd-Sourced Issue Data to Improve User Stories with CrUISE-AC".

domain_thesaurus.json: basic manually created thesaurus for terms that are used in the eCommerce domain. Words are POS tagged ($nfornouns,$v for verbs and $a for adjectives). Composites are combined by underscore. 

Generated Acceptance Criteria.xlsx: automatically generated and assessed acceptance criteria along with a manual approval. The columns are 

• ID: unique ID of the user story
• Connextra: user story written in connextra pattern "As a [role], I [what], {so I [benefit]}"
• Acceptance Criteria: Acceptance criteria that supported the user story originally
• Metric class: class of user story determined by CrUISE-score metric
• IssueID: unique ID of an issue, the acceptance criteria was generated from
• Issue Text Preprocessed: the issue text after beeing preprocessed is supposed to contain requirement-relevant information only
• Generated Acceptance Criteria: Gherkin-style acceptance criteria generated from [Issue Text Preprocessed]
• Explanation: Explanation generated by GPT-4 Turbo why this acceptance criteria is a valuable and non-trivial addition to the user story and its original AC
• Relevant:
  • 1 if manual inspection approved this acceptance criteria as valuable and non-trivial,
  • otherwise 0