Filtered by themeTask performance8 papersclear ✕

Predicting the Need for XAI from High-Granularity Interaction DataVagner Figueredo de Santana, Ana Fucs et al. · 2023Recent advances in Artificial Intelligence (AI) and Machine Learning (ML) brought light on the need for explainability in multiple domains (e.g., healthcare, finance, justice, and recruiting). Explainability or Explainable AI (XAI) can be defined as everything that makes AI more understandable to human beings. However, XAI features may vary according to the AI algorithm used. Beyond XAI features, different AI algorithms vary in terms of speed, performance, and costs associated with training/running models. Knowing when to choose the right algorithm for the task at hand, therefore, is fundamental in multiple AI systems, for instance, AutoML and AutoAI. In this paper, we propose a method to analyze patterns of high-granularity user interface (UI) events (i.e., mouse, keyboard, and additional custom events triggered on the millisecond scale) to predict when users will interact with UI elements that provide explainability for the AI in place. In this context, this paper presents: (1) a user study involving 37 participants (7 in the pilot phase and 30 in the main experiment phase) in which people performed a task of reporting a bug using a text form associated with an AI data quality meter and its XAI UI element and (2) an approach to model micro behavior using 𝑛𝑜𝑑𝑒2𝑣𝑒𝑐 to predict when the interaction with XAI UI element will occur. The proposed approach uses a rich dataset (approximately 129k events) and combines 𝑛𝑜𝑑𝑒2𝑣𝑒𝑐and a Logistic Regression classifier. Results obtained show we have obtained an event-by-event prediction of the interaction with XAI with an average F-score of 0.90 (𝜎= 0.06). From the presented results, one expects to support researchers in the realm of UI personalization to consider high-granularity interaction data when predicting the need for XAI while users are interacting with AI model outputs.

UX of AI & PersonalizationResearch Papers

Combinatorial Optimization of Graphical User Interface DesignsAntti Oulasvirt · 2020| The graphical user interface (GUI) has become the prime means for interacting with computing systems. It lever- ages human perceptual and motor capabilities for elementary tasks such as command exploration and invocation, informa- tion search, and multitasking. For designing a GUI, numerous interconnected decisions must be made such that the out- come strikes a balance between human factors and technical objectives. Normally, design choices are specified manually and coded within the software by professional designers and developers. This article surveys combinatorial optimization as a flexible and powerful tool for computational generation and adaptation of GUIs. As recently as 15 years ago, applications were limited to keyboards and widget layouts. The obstacle has been the mathematical definition of design tasks, on the one hand, and the lack of objective functions that capture essential aspects of human behavior, on the other. This article presents definitions of layout design problems as integer programming tasks, a coherent formalism that permits identification of problem types, analysis of their complexity, and exploitation of known algorithmic solutions. It then surveys advances in Manuscript received April 8, 2019; revised October 9, 2019 and January 12, 2020; accepted January 17, 2020. This work was supported in part by the European Research Council (ERC) through the European Union’s Horizon 2020 Research and Innovation Program under Grant 637991 and in part by the Academy of Finland projects Bayesian Artefact Design (BAD) and Human Automata. (Corresponding author: Antti Oulasvirta.) Antti Oulasvirta is with the Department of Communications and Networking, School of Electrical Engineering, Aalto University, 02150 Espoo, Finland, and also with the Finnish Center for Artificial Intelligence (FCAI), 02015 Espoo, Finland (e-mail: antti.oulasvirta@aalto.fi). Niraj Ramesh Dayama and Morteza Shiripour are with the Department of Communications and Networking, School of Electrical Engineering, Aalto University, 02150 Espoo, Finland. Maximilian John and Andreas Karrenbauer are with the Max Planck Institute for Informatics, 66123 Saarbrücken, Germany. Digital Object Identifier 10.1109/JPROC.2020.2969687 formulating evaluative functions for common design-goal foci such as user performance and experience. The convergence of these two advances has expanded the range of solvable problems. Approaches to practical deployment are outlined with a wide spectrum of applica

UX of AI & PersonalizationResearch Papers

Retroactive Transfer Phenomena inAlternating User Interfaces · 2020We investigated retroactive transfer when users alternate be­ tween different interfaces. Retroactive transfer is the influence of a newly learned interface on users’ performance with a pre­ viously learned interface. In an interview study, participants described their experiences when alternating between different interfaces, e.g. different operating systems, devices or tech­ niques. Negative retroactive transfer related to text entry was the most frequently reported incident. We then reported a lab­ oratory experiment that investigated the impact of similarity between two abstract keyboard layouts, and the number of alternations between them, on retroactive interference. Results indicated that even small changes in the interference interface produced a significant performance drop for the entire previ­ ously learned interface. The amplitude of this performance drop decreases with the number of alternations. We suggest that retroactive transfer should receive more attention in HCI, as the ubiquitous nature of interactions across applications and systems requires users to increasingly alternate between similar interfaces. Author Keywords skill transfer; retroactive interference; keyboard layout CCS Concepts •Human-centered computing → User studies;

UX of AI & PersonalizationResearch Papers

A Long-Term Evaluation of Adaptive Interface Design for Mobile Transit InformationAlexander Haig · 2020Personalization of user experience has a long history of success in the HCI community. More recently the community has focused on adaptive user interfaces, supported by machine learning, that reduce interaction efforts and improves user experience by collaps- ing transactions and pre-filtering results. However, generally, these more recent results have only been demonstrated in the laboratory environment. In this paper, we share the case of a deployed mobile transit app that adapts based on users’ previous usage. We examine the impact of adaptation, both good and bad, and user abandon- ment rates. We conducted an 18-month assessment where 2,616 participants (with and without vision impairments) were recruited and participated in an A/B study. Finally, we draw some insights on some unusual effects that appear over the long term. CCS CONCEPTS • Human-centered computing →Empirical studies in inter- action design.

UX of AI & PersonalizationResearch Papers

Self-adaptation of Workflow Business Software to the User's Requirements and BehaviorUser Interface, User Experience et al. · 2020The main goal of the presented paper is to propose a method for adaptation of the user interface of workflow software to increase its efficiency, reduce the number of errors, and improve its UX. The authors assumed that the system adaptation will be achieved by application to intelligent methods for modeling user as well as system. In order to do this, a special tool for data gathering has been designed and in the next steps of the research, this tool will be also implemented in a real environment. A unique value of the paper is that after many years of theoretical research, the first attempt to implement a practical solution for self-adapting and the personalized interface for workflow systems was done. Janusz Sobecki et al. / Procedia Computer Science 176 (2020) 3506–3513 3507 adaptation [3]. This AI approach for user interface adaptation may be also enhanced with the application of user interface ontologies [4]. 1.1. Related Works In user interface design we should start with proper user model [1] however, we should always remember that ever-increasing number of users, especially of web-based systems, also brings the increase of differences among their users and interaction styles [3]. The user differences may reflex their demographical, psychological as well as sociological user characteristics, which have an influence on the users' information needs and interaction habits. The consequence of these differences is causing difficulties in modeling these users in the standard way [1], so since many years more sophisticated solutions have been proposed, for example, one enhanced with the application of ontologies [5], which was further applied in SOA systems development [4] and [6]. The information systems design and development have been enhanced by ontologies on at least several different levels, such as database integration, business logic, or Graphical User Interfaces (GUI) [6]. The before mentioned work presents an approach for mapping formal ontologies to GUI. This supports device-independent GUI construction and semi-automatic GUI modeling. This issue was also been raised in other work [7], as well as [8]. User interfaces have been also adopted by means of application different recommender methods such as Demographic Filtering, Content-Based Filtering, Collaborative Filtering or Hybrid Approach [9, 15], wherein user grouping or classification different machine learning algorithms may be applied, or the recommender method hybridization may be based

UX of AI & PersonalizationResearch Papers

Individualising Graphical Layouts with Predictive Visual Search ModelsKRIS LUYTEN · 2019In domains where users are exposed to large variations in visuo-spatial features among designs, they often spend excess time searching for common elements (features) on an interface. This article contributes individualised predictive models of visual search, and a computational approach to restructure graphical layouts for an individual user such that features on a new, unvisited interface can be found quicker. It explores four technical principles inspired by the human visual system (HVS) to predict expected positions of features and create individualised layout templates: (I) the interface with highest frequency is chosen as the template; (II) the interface with highest predicted recall probability (serial position curve) is chosen as the template; (III) the most probable locations for features across interfaces are chosen (visual statistical learning) to generate the template; (IV) based on a generative cognitive model, the most likely visual search locations for features are chosen (visual sampling modelling) to generate the template. Given a history of previously seen interfaces, we restructure the spatial layout of a new (unseen) interface with the goal of making its features more easily findable. The four HVS principles are implemented in Familiariser, a web browser that automatically restructures webpage layouts based on the visual history of the user. Evaluation of Familiariser (using visual statistical learning) with users provides first evidence that our approach reduces visual search time by over 10%, and number of eye-gaze fixations by over 20%, during web browsing tasks.

UX of AI & PersonalizationResearch Papers

Predictability and Accuracy in Adaptive User InterfacesKrzysztof Z. Gajos, Katherine Everitt et al. · 2008While proponents of adaptive user interfaces tout potential performance gains, critics argue that adaptation's unpredict- ability may disorient users, causing more harm than good. We present a study that examines the relative effects of predictability and accuracy on the usability of adaptive UIs. Our results show that increasing predictability and accuracy led to strongly improved satisfaction. Increasing accuracy also resulted in improved performance and higher utiliza- tion of the adaptive interface. Contrary to our expectations, improvement in accuracy had a stronger effect on perform- ance, utilization and some satisfaction ratings than the im- provement in predictability. AUTHOR KEYWORDS Adaptive interfaces, predictability, accuracy, user study ACM Classification Keywords H5.2 Information Interfaces and Presentation: User Inter- faces – Interactions Styles, Evaluation/Methodology

UX of AI & PersonalizationResearch Papers

User Interface Design with Combinatorial Optimization1977ptimization methods in user interface (UI) design is a long-standing topic in human-computer interaction (HCI) research. Currently, user-centered design is largely focused on human creativity, sensemaking, empathy, and creation of meaning, but optimization methods have been explored as supplemental ways to help speed up the design cycle and improve design quality (for more information about computational UI design, see the sidebar). Unlike any other design method in HCI, optimization methods offer a greater-than-zero chance of finding an optimal design, and some exact methods even offer mathematical guarantees for solutions. More importantly, if the design task and its

UX of AI & PersonalizationResearch Papers