Extracting Domain Models from User Stories

General Information

• Authors:
  • Sathurshan Arulmohan, McMaster University & McSCert
  • Dr. Marie-Jean Meurs, Université du Québec à Montréal (UQAM) & CIRST
  • Dr. Sébastien Mosser, McMaster University & McSCert
• Version: 1.0 (July 2023)

Content of the repository

• _raw:
  • dalpiaz2018 contains the initial user stories text file from Dalpiaz et al.
  • docanno contains the raw exportation of the annotated dataset, using Docanno's internal format (JSONL).
• ground_truth:
  • Each json file represents the associated anotated backlog, annotated by the 1st and 3rd authors.
• tools:
  • visual_narrator: Concepts extracted using Visual Narrator.
  • gpt-3.5-turbo-0613: Concepts extracted using GPT-3.5, between June 22 and June 24 (2023).
  • crf: concepts extracted using a tailored Conditional Random Fields approach.

File Format

Description

{
  "PID": "backlog identifier",
  "Text": "original text",
  "Persona": [ "..." ],
  "Action": {
    "Primary Action":   [ "...", ... ],
    "Secondary Action": ["...", ...] 
  },
  "Entity": { 
    "Primary Entity":   [ "...", ... ], 
    "Secondary Entity": [ "...", ...] 
  },
  "Benefit": "...",
  "Triggers": [ [ "Persona", "Action" ], ... ],
  "Targets":  [ [ "Action", "Entity" ],  ... ],
  "Contains": [ [ "Entity", "Entity" ],  ... ]
}

Example

{
    "PID": "#G02#",
    "Text": "#G02# As a UI designer, I want to begin user testing, so that I can validate stakeholder UI improvement requests.",
    "Persona": ["UI designer"],
    "Action": {
        "Primary Action": ["begin"],
        "Secondary Action": ["validate"]
    },
    "Entity": {
        "Primary Entity": ["user testing"],
        "Secondary Entity": ["stakeholder UI improvement requests"]
    },
    "Benefit": "I can validate stakeholder UI improvement requests",
    "Triggers": [ ["UI designer", "begin"] ],
    "Targets": [
        ["begin", "user testing"],
        ["validate", "stakeholder UI improvement requests"]
    ],
    "Contains": []
}

Publications using this repository:

Published

1. Modelling Agile Backlogs as Composable Artefacts to support Developers and Product Owners. Sébastien Mosser, Vladimir Reihnarz, and Corinne Pulgar. Journal of Object Technology, 2022. (www)

Submitted

1. Extracting Domain Models from Textual Requirements in the Era of Large Language Models. Sathurshan Arulmohan, Marie-Jean Meurs, and Sébastien Mosser. Submitted to the 5th Workshop on Artificial Intelligence and Model-driven Engineering, co-located with MODELS'2023.