Crying Jaywalker! Notifying Take-Over-Requests and Critical Events in Operational Driving Domain of Autonomous Vehicles via Multimodal InterfacesThe advent of self-driving cars promises to enable occupants to repurpose commuting time. However, although in conditional automation (SAE Level 3) drivers can engage in non-driving related tasks (NDRTs), they must be ready to intervene when prompted by the system with a take-over request (TOR). The vehicle may also need to warn the driver about critical events without notifying a TOR (as in sudden hard braking due to a jaywalker). Clearly and effectively communicating these events and their urgency is crucial for the successful adoption of autonomous vehicles. This work analyzes the impact of multimodal visual and audio cues in conveying this information. It considers an augmented reality (AR) windshield display (WSD) combining screen-fixed elements and world-registered AR overlays, alongside an auditory interface providing explanations and alerts through speech and abstract sounds. The effectiveness of these combined stimuli was evaluated through a user study conducted in a VR-based driving simulator.2025FPFilippo Gabriele Pratticò et al.Automated Driving Interface & Takeover DesignHead-Up Display (HUD) & Advanced Driver Assistance Systems (ADAS)IUI
DeepFlow: A Flow-Based Visual Programming Tool for Deep Learning DevelopmentVisual programming tools have recently been introduced to enable Deep Learning (DL) development without the need for expertise in traditional programming languages and frameworks. However, these tools often exhibit limitations in scalability for complex architectures and lack real-time debugging capabilities. This paper introduces DeepFlow, a flow-based visual programming tool (VPT) designed to address these challenges by leveraging the inherently visual nature of DL models as sequences of learnable functions. DeepFlow incorporates hierarchical abstraction mechanisms through ``supernodes'' to support model scalability, which is crucial for modern, complex architectures. Additionally, it introduces interactive debugging in the model design phase, allowing developers to validate network architectures before execution. We conducted a user study with 16 DL developers, involving typical DL model design tasks. We assessed DeepFlow using quantitative usability metrics, and post-task interviews to evaluate user perceptions and workflow integration across different expertise levels. Results demonstrated high usability and user satisfaction, and highlighted DeepFlow's effectiveness for rapid model iteration and as a learning aid for complex DL architectures, while also identifying areas for improvement.2025TCTommaso Calò et al.Interactive Data VisualizationComputational Methods in HCIIUI
Dramatic Things: Investigating Value Conflicts in Smart Home through Enactment and Co-speculationSmart home technologies embed values such as sustainability, comfort, privacy, and security, which can sometimes conflict with one another, considering the complexities of domestic environments. This paper investigates the potential implications of these value conflicts and the corresponding design challenges. Through an enactment session and co-speculations with professional actors, we explored what it means to navigate multiple values simultaneously, live with products that impose their own values, and manage value conflicts both with and among smart products. The findings challenge the seamless and harmonious vision of smart homes conceived by technologists, proposing shifts in the common narrative: from value alignment to value transparency, from service provision to mutual care, and from autonomy to responsiveness. We discuss that acknowledging value conflicts, rather than eliminating them, is an opportunity to gain a deeper understanding of users and home environments and guide the design of smart home technologies.2025NCNazli Cila et al.Delft University of Technology, Human-Centered DesignContext-Aware ComputingHome Energy ManagementSmart Home Privacy & SecurityCHI
Investigating How Computer Science Researchers Design Their Co-Writing Experiences With AIRecent advancements in AI have significantly enhanced collaboration between humans and writing assistants. However, empirical evidence is still lacking on how this collaboration unfolds in scientific writing, especially considering the variety of tools researchers can use nowadays. We conducted observations and retrospective interviews to investigate how 19 computer science researchers collaborated with intelligent writing assistants while working on their ongoing projects. We adopted a design-in-use lens to analyze the collected data, exploring how researchers adapt writing assistants during their use to overcome challenges and meet their specific needs and preferences. Our findings identify issues such as workflow disruptions and over-reliance on AI, and reveal five distinct design-in-use styles---teaching, resisting, repurposing, orchestrating, and complying---each consisting of different practices used by researchers. This study contributes to understanding the evolving landscape of human-AI co-writing in scientific research and offers insights for designing more effective writing assistants.2025ARAlberto Monge Roffarello et al.Politecnico di Torino, Dipartimento di Automatica e InformaticaHuman-LLM CollaborationCreative Collaboration & Feedback SystemsCHI
NLPGuard: A Framework for Mitigating the Use of Protected Attributes by NLP ClassifiersAI regulations are expected to prohibit machine learning models from using sensitive attributes during training. However, the latest Natural Language Processing (NLP) classifiers, which rely on deep learning, operate as black-box systems, complicating the detection and remediation of such misuse. Traditional bias mitigation methods in NLP aim for comparable performance across different groups based on attributes like gender or race but fail to address the underlying issue of reliance on protected attributes. To partly fix that, we introduce NLPGuard, a framework for mitigating the reliance on protected attributes in NLP classifiers. NLPGuard takes an unlabeled dataset, an existing NLP classifier, and its training data as input, producing a modified training dataset that significantly reduces dependence on protected attributes without compromising accuracy. NLPGuard is applied to three classification tasks: identifying toxic language, sentiment analysis, and occupation classification. Our evaluation shows that current NLP classifiers heavily depend on protected attributes, with up to 23% of the most predictive words associated with these attributes. However, NLPGuard effectively reduces this reliance by up to 79%, while slightly improving accuracy.2024SGSalvatore Greco et al.Session 3c: Evolving Approaches to PrivacyCSCW
Defining and Identifying Attention Capture Deceptive Designs in Digital InterfacesMany tech companies exploit psychological vulnerabilities to design digital interfaces that maximize the frequency and duration of user visits. Consequently, users often report feeling dissatisfied with time spent on such services. Prior work has developed typologies of damaging design patterns (or dark patterns) that contribute to financial and privacy harms, which has helped designers to resist these patterns and policymakers to regulate them. However, we are missing a collection of similar problematic patterns that lead to attentional harms. To close this gap, we conducted a systematic literature review for what we call 'attention capture damaging patterns' (ACDPs). We analyzed 43 papers to identify their characteristics, the psychological vulnerabilities they exploit, and their impact on digital wellbeing. We propose a definition of ACDPs and identify eleven common types, from Time Fog to Infinite Scroll. Our typology offers technologists and policymakers a common reference to advocate, design, and regulate against attentional harms.2023ARAlberto Monge Roffarello et al.Politecnico di TorinoDark Patterns RecognitionCHI
TiiS: Understanding, Discovering, and Mitigating Habitual Smartphone Use in Young AdultsRoffarello 等人针对年轻人群的智能手机使用习惯进行研究，旨在理解、发现并缓解其中的问题性使用行为。2022ARAlberto Monge Roffarello et al.Voice User Interface (VUI) DesignSocial Platform Design & User BehaviorIUI
HandPainter --- 3D Sketching in VR with Hand-based Physical Proxy3D sketching in virtual reality (VR) enables users to create 3D virtual objects intuitively and immersively. However, previous studies showed that mid-air drawing may lead to inaccurate sketches. To address this issue, we propose to use one hand as a canvas proxy and the index finger of the other hand as a 3D pen. To this end, we first perform a formative study to compare two-handed interaction with tablet-pen interaction for VR sketching. Based on the findings of this study, we design HandPainter, a VR sketching system which focuses on the direct use of two hands for 3D sketching without requesting any tablet, pen, or VR controller. Our implementation is based on a pair of VR gloves, which provide hand tracking and gesture capture. We devise a set of intuitive gestures to control various functionalities required during 3D sketching, such as canvas panning and drawing positioning. We show the effectiveness of HandPainter by presenting a number of sketching results and discussing the outcomes of a user study-based comparison with mid-air drawing and tablet-based sketching tools.2021YJYing Jiang et al.The University of Hong KongHand Gesture RecognitionFull-Body Interaction & Embodied Input3D Modeling & AnimationCHI
Coping with Digital Wellbeing in a Multi-Device WorldWhile Digital Self-Control Tools (DSCTs) mainly target smartphones, more effort should be put into evaluating multi-device ecosystems to enhance digital wellbeing as users typically use multiple devices at a time. In this paper, we first review more than 300 DSCTs by demonstrating that the majority of them implements a single-device conceptualization that poorly adapts to multi-device settings. Then, we report on the results from an interview and a sketching exercise (N=20) exploring how users make sense of their multi-device digital wellbeing. Findings show that digital wellbeing issues extend beyond smartphones, with the most problematic behaviors deriving from the simultaneous usage of different devices to perform uncorrelated tasks. While this suggests the need of DSCTs that can adapt to different and multiple devices, our work also highlights the importance of learning how to properly behave with technology, e.g., through educational courses, which may be more effective than any lock-out mechanism.2021ARAlberto Monge Roffarello et al.Politecnico di TorinoContext-Aware ComputingNotification & Interruption ManagementWorkplace Wellbeing & Work StressCHI
Empowering End Users in Debugging Trigger-Action RulesEnd users can program trigger-action rules to personalize the joint behavior of their smart devices and online services. Trigger-action programming is, however, a complex task for non-programmers and errors made during the composition of rules may lead to unpredictable behaviors and security issues, e.g., a lamp that is continuously flashing or a door that is unexpectedly unlocked. In this paper, we introduce EUDebug, a system that enables end users to debug trigger-action rules. With EUDebug, users compose rules in a web-based application like IFTTT. EUDebug highlights possible problems that the set of all defined rules may generate and allows their step-by-step simulation. Under the hood, a hybrid Semantic Colored Petri Net (SCPN) models, checks, and simulates trigger-action rules and their interactions. An exploratory study on 15 end users shows that EUDebug helps identifying and understanding problems in trigger-action rules, which are not easily discoverable in existing platforms.2019FCFulvio Corno et al.Politecnico di TorinoSmart Home Interaction DesignPrototyping & User TestingCHI
The Race Towards Digital Wellbeing: Issues and OpportunitiesAs smartphone use increases dramatically, so do studies about technology overuse. Many different mobile apps for breaking "smartphone addiction" and achieving "digital wellbeing" are available. However, it is still not clear whether and how such solutions work. Which functionality do they have? Are they effective and appreciated? Do they have a relevant impact on users' behavior? To answer these questions, (i) we reviewed the features of 42 digital wellbeing apps, (ii) we performed a thematic analysis on 1,128 user reviews of such apps, and (iii) we conducted a 3-week-long in-the-wild study of Socialize, an app that includes the most common digital wellbeing features, with 38 participants. We discovered that digital wellbeing apps are appreciated and useful for some specific situations. However, they do not promote the formation of new habits and they are perceived as not restrictive enough, thus not effectively helping users to change their behavior with smartphones.2019ARAlberto Monge Roffarello et al.Politecnico di TorinoMental Health Apps & Online Support CommunitiesCHI
WeBrowse: Leveraging User Clicks for Content Discovery in Communities of a PlaceOne of the limits of web content discovery tools, let them be recommender systems or content curation tools such as social rating, social bookmarking and other social media, is the scarcity of user input (e.g. rate, submit, share). This problem is even worse in the case of what we call communities of a place: people who study, live or work at the same place. Such people often share common interests but either do not know each other or fail to actively engage in submitting and relaying information. In this paper, we investigate the feasibility of using the aggregated clicks of entire communities of users to passively emulate a content curation service a la Reddit. To this end, we prototype and deploy WeBrowse, a content curation service based on the processing of raw HTTP logs. Evaluation based on our deployments demonstrates feasibility at scale while respecting user privacy. The majority of WeBrowse's users welcome the quality of content it promotes.2018GSGiuseppe Scavo et al.Crowdsourcing SystemsCSCW