Classifying design‐level requirements using machine learning for a recommender of interaction design patterns

Abstract

Software Engineering is a discipline that encompasses processes associated with the development of interactive systems. The perceived quality of an interactive system is heavily influenced by the user interface design, which may result in many challenges. One such challenge is design‐level requirements analysis. The success of the software system is mostly dependent on how well users’ requirements have been understood and translated into appropriate functionalities. During the interactive system design process, it is common to find recurring problems in human–computer interactions, for which reusing solutions is highly feasible. Interaction design patterns seek to support designers in decision making during the design of interactive systems. Due to the design task tends to be subjective and prone to errors. This work aims at presenting and evaluating an interaction design patterns recommendation model based on design‐level requirements classification, through the application of supervised machine learning algorithms. To compare the performance of four classification algorithms, a study was carried out, in which the linear support vector machine was the most suitable to this problem. The results of this work can be used for implementing frameworks that can better support designers’ decision making when designing user interfaces.

Artificial intelligence (AI) for user experience (UX) design: a systematic literature review and future research agenda

“…In fact, cases of bad UX have resulted in non-use by end-users and have been associated with technostress, fatigue, and misuse (Hart & Sutcliffe, 2019;Nisafani et al, 2020). End users of digital solutions have high standards for what they demand of an application or service, and software success is often linked to how well designers manage to understand and translate requirements into corresponding functionality and appropriate aesthetics (Silva-Rodríguez et al, 2020). Following an iterative user centred design process, and having qualities of creativity, problem-solving, sense-making, empathy, and collaboration, is shown to result in user friendly and innovative solutions (Oulasvirta et al, 2020).…”

“…In addition, scholars mostly agree that AI should not be used for automating the entire process but rather offer to designers the tools that can make the design process easier and more accurate (e.g., Feldman, 2017b;Gardey et al, 2022). For instance, supervised machine learning can be used for enabling designers identify design patterns that perform better in terms of usability (Silva-Rodríguez et al, 2020) and deep learning techniques can improve design performance when iterating initial designs (Chaudhuri et al, 2022b;).…”

The Use of NLP-Based Text Representation Techniques to Support Requirement Engineering Tasks: A Systematic Mapping Review

“…Based on this type of representation, each requirement is represented as a vector of words or tokens or stems combined with their weights. These vectors are used in the subsequent phase to apply similarity-based rules [55], [71] or as an input to train machine learning models (such as SVM [40], [69], [87], [141]). The output of both techniques is a k-dimensional space that reflects a k latent topics within the processed requirements.…”

The Use of NLP-Based Text Representation Techniques to Support Requirement Engineering Tasks: A Systematic Mapping Review

“…Based on this type of representation, each requirement is represented as a vector of words or tokens or stems combined with their weights. These vectors are used in the subsequent phase to apply similarity-based rules [55,71] or as an input to train machine learning models (such as SVM [87,69,40,141]).…”