Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Empowering Agile-Based Generative Software Development through Human-AI Teamwork

Sai Zhang, Zhenchang Xing, Ronghui Guo, Fangzhou Xu, Lei Chen, Zhaoyuan Zhang, Xiaowang Zhang, Zhiyong Feng, Zhiqiang Zhuang

TL;DR

By introducing the Gherkin language, AgileGen attempts for the first time to use testable requirement descriptions as a bridge for semantic consistency between requirements and code, aiming to ensure that software products meet actual user requirements by defining user scenarios that include acceptance criteria.

Abstract

In software development, the raw requirements proposed by users are frequently incomplete, which impedes the complete implementation of application functionalities. With the emergence of large language models, recent methods with the top-down waterfall model employ a questioning approach for requirement completion, attempting to explore further user requirements. However, users, constrained by their domain knowledge, lack effective acceptance criteria, which fail to capture the implicit needs of the user. Moreover, the cumulative errors of the waterfall model can lead to discrepancies between the generated code and user requirements. The Agile methodologies reduce cumulative errors through lightweight iteration and collaboration with users, but the challenge lies in ensuring semantic consistency between user requirements and the code generated. We propose AgileGen, an agile-based generative software development through human-AI teamwork. AgileGen attempts for the first time to use testable requirements by Gherkin for semantic consistency between requirements and code. Additionally, we innovate in human-AI teamwork, allowing users to participate in decision-making processes they do well and enhancing the completeness of application functionality. Finally, to improve the reliability of user scenarios, a memory pool mechanism is used to collect user decision-making scenarios and recommend them to new users. AgileGen, as a user-friendly interactive system, significantly outperformed existing best methods by 16.4% and garnered higher user satisfaction.

Empowering Agile-Based Generative Software Development through Human-AI Teamwork

TL;DR

By introducing the Gherkin language, AgileGen attempts for the first time to use testable requirement descriptions as a bridge for semantic consistency between requirements and code, aiming to ensure that software products meet actual user requirements by defining user scenarios that include acceptance criteria.

Abstract

In software development, the raw requirements proposed by users are frequently incomplete, which impedes the complete implementation of application functionalities. With the emergence of large language models, recent methods with the top-down waterfall model employ a questioning approach for requirement completion, attempting to explore further user requirements. However, users, constrained by their domain knowledge, lack effective acceptance criteria, which fail to capture the implicit needs of the user. Moreover, the cumulative errors of the waterfall model can lead to discrepancies between the generated code and user requirements. The Agile methodologies reduce cumulative errors through lightweight iteration and collaboration with users, but the challenge lies in ensuring semantic consistency between user requirements and the code generated. We propose AgileGen, an agile-based generative software development through human-AI teamwork. AgileGen attempts for the first time to use testable requirements by Gherkin for semantic consistency between requirements and code. Additionally, we innovate in human-AI teamwork, allowing users to participate in decision-making processes they do well and enhancing the completeness of application functionality. Finally, to improve the reliability of user scenarios, a memory pool mechanism is used to collect user decision-making scenarios and recommend them to new users. AgileGen, as a user-friendly interactive system, significantly outperformed existing best methods by 16.4% and garnered higher user satisfaction.
Paper Structure (49 sections, 4 equations, 20 figures, 6 tables)

This paper contains 49 sections, 4 equations, 20 figures, 6 tables.

Table of Contents

  1. Introduction
  2. MOTIVATION & Innovation
  3. Motivation
  4. Innovation
  5. Design of Human-AI Collaborative AgileGen
  6. AgileGen Overview
  7. User Decision-Making & Interaction Interface Design
  8. User Decision-Making
  9. AgileGen Interaction Interface Design
  10. Design of Core Components
  11. Scenario Design component
  12. Rapid Prototype Design component.
  13. EXPERIMENTAL SETUP
  14. Research Questions
  15. Datasets
  16. ...and 34 more sections

Figures (20)

  • Figure 1: Problem Statement Diagram. Users are unsure how to drive the Agent to generate desired software, and the Agent does not know how to fulfill user requirements. We have built a bridge between users and the Agent, facilitating collaboration between human decision-making skills and the Agent's coding capabilities. This collaboration has created a generative software development Agent with lightweight iterative feedback.
  • Figure 2: Requirement form & Human aspects on the performance of individuals in Requirements Engineering (RE)-related activities
  • Figure 3: Comparing a software company with an AI-based agent in software development.
  • Figure 4: Display of the code generation effects by AutoGPT, DemoGPT, MetaGPT, and ChatDev agents for the requirement "Please generate a web system code with random roll call function".
  • Figure 5: AgileGen Overview Diagram. The left side displays the user decision-making part, while the right shows the AgileGen Agent's automatic generation. The black arrow in the middle represents the interaction between the user and the Agent. The process starts with the End User requirement decision, followed sequentially by the Scenario Design component, Scenario decision-making, Rapid Prototype Design, and Acceptance & recommendation decision-making. The outcome is the software code.
  • ...and 15 more figures