1. Introduction
Agile development is a prominent research direction in requirements engineering. Recently, novel advancements in theory have emerged, while abundant practical experience has been obtained in various enterprise operations in the Internet and big data environments
[1]. At its core, agile development is a development method that responds to quickly changing user requirements through iterative, small-scale, and rapid development and functionality delivery, ultimately allowing for a rapid response to user requirements and the continuous optimization of function design
[2]. Compared to the traditional waterfall approach, agile development has a significant advantage in dealing with unclear or changing requirements, and can greatly minimize the risk of costly user requirement changes in the later stages of project development
[3].
The successful implementation of agile development in a project relies on the efficient, rapid, and accurate flow of information between organizations or members
[4]. Of all kinds of information, user requirements are of the utmost importance, serving as the core of agile development teams. Inaccurate requirements can impede progress, drastically impact the quality of the delivered product, and even damage the morale of the development team. Consequently, high-quality user requirements can be seen as a “catalyst” that provides a solid foundation in the early stages of a project, bolsters the positive feedback of iterative development, leads the entire project development process in a virtuous cycle, and ultimately results in a product that effectively meets user needs. Requirements play a crucial role in project development across various fields. For example, in research on optical wireless communication systems based on smartphone cameras, the technology requirements of smartphone cameras play a critical and leading role in generating solutions for optical camera communication
[5]. In the field of the model-based design of cyber-physical systems, traceability from requirements to the model to the simulation results has become increasingly important
[6]. In the development of mobile learning applications, inappropriate technology requirements will affect the quality and increase the development cost of mobile learning applications
[7]. Additionally, in the Internet environment, users can add their comments as feedback in the mobile application store, and this feedback can be considered a requirement and analyzed through natural language processing to improve the software quality and functionality of mobile applications
[8][9][8,9].
In agile development, user stories are commonly used as the source of user requirements instead of traditional requirement specifications. A user story is composed of three elements, namely role, activity, and value, and is usually expressed as “What ‘Activity’ a ‘role’ wants to accomplish to achieve what ‘Value’”
[10][11][10,11]. By collecting a large number of user stories, agile development teams can quickly comprehend user requirements, allowing for rapid response and iteration. However, in many project practices, the agile development process with user stories as a tool is not always implemented smoothly
[12][13][12,13]. This is mainly due to the wide source of user requirements, as well as different cultural backgrounds, abilities, and cooperation degrees of users, causing most of the collected user stories to be incomplete, inconsistent, and inaccurate, thereby reducing the usability of user requirement information and seriously affecting the efficiency of agile development.
The traditional approach to these problems involves setting guidelines prior to collecting user stories. The INVEST (independent, negotiable, valuable, estimable, small, testable) principles provide a framework for evaluating the quality of user stories
[14]. Agile teams typically analyze user stories through manual review after collection and by filtering out qualified stories for development. While these strategies improve quality to some extent, they are impractical when the number of user stories is large. Investing too much time and effort in review can burden the agile team, impeding iteration speed and making it difficult to meet requirements on time. Thus, given the pragmatic requirements of agile development practices, it is critical to construct a scientific, automatic, and effective user requirement quality assessment method to assist agile teams in improving requirement quality and development efficiency.
2. Definition of Requirement Quality
Requirement quality refers to the degree to which requirements satisfy the needs of stakeholders and meet the goals of project development. In other words, high-quality requirements are those that clearly and unambiguously describe what the system needs to carry out without any inconsistencies, conflicts, or omissions [15]. Quality requirements are typically characterized by completeness, consistency, correctness, clarity, and testability [16]. They should be able to provide a clear and precise specification of what a system should do, what it should not do, and how it should behave under different conditions [17]. The definition of requirement quality has been given new connotations through the continuous efforts of researchers. Calazans et al. [18] defined low-quality requirements as “requirements smells” and confirmed the existence of requirements smells in the quality requirement specifications classified by ISO/IEC 25010 by analyzing 26 institutional regulations of a large public financial institution. Eugenio et al. [19] extended the definition of requirement quality in the temporal dimension and developed an industrial method for analyzing the evolution of requirement quality, which is implemented in the system quality analyzer tool. Although the definition of requirement quality is constantly being refined and innovated, the applicability of requirement quality definitions has not yet been matched and modified in the field of agile development, and relevant mathematical models and assessment methods for requirement quality have not been established.
3. Assessment of Requirement Quality
Assessing requirement quality is an essential task in project development [20]. Numerous methods and techniques have been proposed for requirement quality assessment. These methods can be broadly classified into two categories: subjective and objective. Subjective methods involve human judgment and expert opinions, such as manual reviews [21][22][21,22], walkthroughs [23][24][23,24], and surveys [24][25][24,25]. Objective methods, on the other hand, employ automated tools and formal techniques, such as static analysis [26], natural language processing [27][28][29][30][27,28,29,30], machine learning [20][31][20,31], and deep learning [32]. Objective methods have gained increasing attention due to their advantages of automation, repeatability, and scalability. Wang et al. [33] proposed an approach to automatically detect defects in requirements. By utilizing structural, syntactic, and semantic analysis, this approach can automatically improve the quality of requirements. Valentin et al. [31] developed a novel and flexible approach for assessing requirement quality based on machine learning techniques that can be adapted to various environments, projects, organizations, and quality standards. Gregorius et al. [32] utilized deep learning methods to investigate the impact of diverse software domains on quality attribute prediction. This approach can assist users in identifying quality attributes in their requirement writing. Although these objective methods have improved the efficiency of requirement quality assessment, the evaluation results of requirement development teams are still important indicators that cannot be ignored in agile development practices. Therefore, it is necessary to combine subjective and objective requirement evaluation methods to improve the applicability and effectiveness of addressing requirement quality issues in agile development.
4. Relationship with Project Development Outcomes
Requirement quality has a significant impact on project development outcomes, such as development costs, user satisfaction, and project success
[30][34][35][30,34,35]. Low-quality requirements can lead to project development failures, project delays, cost overruns, and customer dissatisfaction
[25][36][25,36]. High-quality requirements, on the other hand, can facilitate the development of systems that meet user needs, are easy to use, and have good performance and reliability
[26][37][26,37]. Therefore, ensuring requirement quality is essential to achieving project development success. Eman et al.
[25] conducted a survey of Egyptian software development practitioners through questionnaires and revealed the association between requirement quality and software development success rates. Furthermore, Amalinda et al.
[30] evaluated the efficacy of IBM’s requirement quality assistant tool in aiding the requirement quality assessment of Bosch requirement engineers, and suggested applicable improvement recommendations. Emil et al.
[36] interviewed 20 project practitioners, including requirement engineers, developers, and testers, and found that when they dealt with requirements they considered to be of low quality, they experienced negative emotions, more work, and additional communication, which indirectly led to problems such as slow project progress and low completion quality of development tasks. As seen, requirement quality is a key factor that influences project development outcomes. Researchers in agile development also need to focus on requirement quality issues, and targeted requirement quality assessment methods need to be proposed.
In general, there exists a plethora of research on requirement quality in requirements engineering, yet few studies have focused on modeling and analyzing the user requirement quality assessment process in agile development. Agile development is distinct from other requirement engineering approaches because it emphasizes faster response and iteration of user requirements and functional design, thereby necessitating higher requirement quality. Drawing on the extant research, rethisearchers paper proposes a user requirement quality assessment process and constructs a user requirement quality assessment method, which is a pioneering and significant endeavor in agile requirement engineering.
5. User Requirement Quality Assessment Framework
The agile development process is illustrated in
Figure 1, which is extended with the overall framework of requirement quality assessment
in this paper. When the requirements are not clearly specified, the project team should start by identifying target users and forming a target user group. Subsequently, user story cards are distributed to the target user group, and the purpose and standards of filling in the cards are explained to them. After the user story cards are filled out, they are collected back by the project team. Then, the quality of user requirements is automatically evaluated by computers, and only qualified user requirements are fed into fast development. Finally, the product or function is delivered to the user, and the rapid iteration of product development is achieved through user feedback.
Figure 1. Framework for user requirement quality assessment in agile development.
As illustrated in
Figure 1, the quality assessment of user requirements is conducted from two perspectives: individual user requirements and overall user requirements. Based on seven quality assessment criteria, the process is divided into three steps. First, an individual user requirement is assessed from four aspects: complete, consistent, simple, and accurate. Second, unqualified requirements are fed back to the users and modified. Finally, the overall quality assessment of all requirements with partially qualified quality is conducted from three aspects: robust, unique, and harmonious. Fully qualified user requirements are generated by adding, deleting, and modifying requirements.
The seven assessment criteria are further explicated in the following section.
Encyclopedia
