Requirements for Active Assistance of Natural Questions in Software Architecture

TL;DR

This paper explores the lifecycle and requirements for an AI-supported environment to manage natural questions during software architecture design, aiming to improve knowledge preservation and decision-making.

Contribution

It proposes a lifecycle model for natural questions and identifies key functional and non-functional requirements for an AI-enhanced support environment.

Findings

Validated requirements through expert survey

Identified key challenges in managing natural questions

Proposed an adaptable, knowledge-integrated support environment

Abstract

Natural questions are crucial to shaping key architectural decisions and preserving architectural knowledge. They arise organically during the architectural design process, often resulting from the existing architectural experience of the designer and the distinctive characteristics of the system being designed. However, natural questions are often mismanaged or ignored, which can lead to architectural drift, knowledge loss, inefficient resource use, or poor understandability of the system's architecture. We aim to better understand the lifecycle of natural questions, its key requirements, challenges and difficulties, and then to envision an assisted environment to properly support it. The environment should be adaptable and responsive to real-world constraints and uncertainties by seamlessly integrating knowledge management tools and artificial intelligence techniques into software development workflows. Based on existing literature, a requirements workshop, and three design iterations, we proposed a lifecycle for natural questions and elicited essential functional and non-functional requirements for such an environment. At last, the results of a survey conducted with experts helped to analyze and validate the elicited requirements and proposed features for the environment to enhance collaboration, decision-making, and the preservation of architectural knowledge more effectively than conventional methods.