SCSimulator: An Exploratory Visual Analytics Framework for Partner Selection in Supply Chains through LLM-driven Multi-Agent SimulationSupply chains (SCs), complex networks spanning from raw material acquisition to product delivery, with enterprises as interconnected nodes, play a pivotal role in organizational success. However, optimizing SCs remains challenging, particularly in partner selection, a key bottleneck shaped by both competitive and cooperative dynamics. This challenge inherently constitutes a multi-objective dynamic game requiring a synergistic integration of Multi-Criteria Decision-Making (MCDM) and Game Theory (GT). Traditional approaches, grounded in mathematical simplifications and managerial heuristics, often fail to capture real-world intricacies and risk introducing subjective biases. Multi-agent simulation (MAS) offers promise, but prior research has largely relied on fixed, uniform agent logic, limiting practical applicability. Recent advances in Large Language Models (LLMs) create new opportunities to represent complex SC requirements and hybrid game logic. However, challenges persist in modeling dynamic SC relationships, ensuring interpretability, and balancing agent autonomy with expert control. To address these issues, we present SCSimulator, an exploratory visual analytics framework that integrates LLM-driven MAS with human-in-the-loop collaboration for SC partner selection. SCSimulator simulates SC evolution via adaptive network structures and enterprise behaviors, which are visualized via interpretable interfaces. By combining Chain-of-Thought (CoT) reasoning with explainable AI (XAI) techniques, the framework generates multi-faceted, transparent explanations of decision trade-offs. Users can iteratively adjust simulation settings to explore outcomes aligned with their expectations and strategic priorities. Developed through iterative co-design with SC experts and industry managers, SCSimulator serves as a proof-of-concept, offering both methodological contributions and practical insights for future research on SC decision-making and interactive AI-driven analytics. Usage scenarios and a user study further demonstrate the system's effectiveness and usability.2026SGShenghan Gao et al.ShanghaiTech UniversityHuman-LLM CollaborationExplainable AI (XAI)Interactive Data VisualizationIUI
From Platform Data to Personal Insight: How Users Make Sense of and Reflect on Personalized Social Media Annual RecapsSocial media platforms generate personalized annual recaps presenting algorithmically-curated summaries of users’ online activities. Unlike traditional personal informatics where users actively collect data, these recaps present unsolicited insights demanding sensemaking effort. Through interviews with 20 participants and analysis of annual recaps, we investigated how users make sense of and reflect on these presentations. We identified seven data presentation types and five sensemaking activities facilitating different reflection levels. We found that concrete presentations like extreme details serve as foundational anchors across all levels, while more abstract presentations predominantly prompt critical reflection. Sensemaking activities lead to reflection through four paths: descriptive reflection involves scanning and annotation, dialogic reflection requires explanation-seeking activities, transformative reflection involves comprehensive sensemaking processes with emphasis on verification, while critical reflection can emerge from any path. We contribute theoretical bridges between sensemaking and reflection in personal informatics and provide design implications to support sensemaking and reflection on personal data.2026WLWenqi Li et al.Peking UniversityBehavior Change & Reflection TechnologyData-Driven Personal Decision-MakingSocial Platform Design & User BehaviorCHI
Augmenting Clinical Decision-Making with an Interactive and Interpretable AI Copilot: A Real-World User Study with Clinicians in Nephrology and ObstetricsClinician skepticism toward opaque AI hinders adoption in high-stakes healthcare. We present AICare, an interactive and interpretable AI copilot for collaborative clinical decision-making. By analyzing longitudinal electronic health records, AICare grounds dynamic risk predictions in scrutable visualizations and LLM-driven diagnostic recommendations. Through a within-subjects counterbalanced study with 16 clinicians across nephrology and obstetrics, we comprehensively evaluated AICare using objective measures (task completion time and error rate), subjective assessments (NASA-TLX, SUS, and confidence ratings), and semi-structured interviews. Our findings indicate AICare's reduced cognitive workload. Beyond performance metrics, qualitative analysis reveals that trust is actively constructed through verification, with interaction strategies diverging by expertise: junior clinicians used the system as cognitive scaffolding to structure their analysis, while experts engaged in adversarial verification to challenge the AI's logic. This work offers design implications for creating AI systems that function as transparent partners, accommodating diverse reasoning styles to augment rather than replace clinical judgment.2026YZYinghao Zhu et al.Peking UniversityExplainable AI (XAI)AI-Assisted Decision-Making & AutomationEV Charging & Eco-Driving InterfacesCHI
NarrativeLoom: Enhancing Creative Storytelling through Multi-Persona Collaborative ImprovisationLarge Language Models show promise for AI-assisted storytelling, yet current tools often generate predictable, unoriginal narratives. To address this limitation, we present NarrativeLoom, a multi-persona co-creative system grounded in Campbell's Blind Variation and Selective Retention theory. NarrativeLoom deploys specialized Artificial Intelligence (AI) personas to generate diverse narrative options (blind variation), while users act as creative directors to select and refine them (selective retention). We designed a controlled study with 50 participants and found that stories co-authored with NarrativeLoom were not only perceived by users as more novel and diverse but were also objectively rated by experts as significantly better across all Torrance Test creativity dimensions: fluency, flexibility, originality, and elaboration. Stories are significantly longer with richer settings and more dialogue. Writing expertise emerged as a moderator: novices benefited more from structured scaffolding. This demonstrates the value of theory-informed co-creative systems and the importance of adapting them to varying user expertise. Project page: https://ppyyqq.github.io/narrativeloom.2026YMYuxi Ma et al.Peking UniversityGenerative AI (Text, Image, Music, Video)Human-LLM CollaborationAI-Assisted Creative WritingCHI
ZuantuSet: A Collection of Historical Chinese Visualizations and IllustrationsHistorical visualizations are a valuable resource for studying the history of visualization and inspecting the cultural context where they were created. When investigating historical visualizations, it is essential to consider contributions from different cultural frameworks to gain a comprehensive understanding. While there is extensive research on historical visualizations within the European cultural framework, this work shifts the focus to ancient China, a cultural context that remains underexplored by visualization researchers. To this aim, we propose a semi-automatic pipeline to collect, extract, and label historical Chinese visualizations. Through the pipeline, we curate ZuantuSet, a dataset with over 71K visualizations and 108K illustrations. We analyze distinctive design patterns of historical Chinese visualizations and their potential causes within the context of Chinese history and culture. We illustrate potential usage scenarios for this dataset, summarize the unique challenges and solutions associated with collecting historical Chinese visualizations, and outline future research directions.2025XMXiyao Mei et al.Peking University, National Key Laboratory of General Artificial Intelligence and School of Intelligence Science and TechnologyInteractive Data VisualizationMuseum & Cultural Heritage DigitizationCHI
Beyond Explicit and Implicit: How Users Provide Feedback to Shape Personalized Recommendation ContentAs personalized recommendation algorithms become integral to social media platforms, users are increasingly aware of their ability to influence recommendation content. However, limited research has explored how users provide feedback through their behaviors and platform mechanisms to shape the recommendation content. We conducted semi-structured interviews with 34 active users of algorithmic-driven social media platforms (e.g., Xiaohongshu, Douyin). In addition to explicit and implicit feedback, this study introduced intentional implicit feedback, highlighting the actions users intentionally took to refine recommendation content through perceived feedback mechanisms. Additionally, choices of feedback behaviors were found to align with specific purposes. Explicit feedback was primarily used for feed customization, while unintentional implicit feedback was more linked to content consumption. Intentional implicit feedback was employed for multiple purposes, particularly in increasing content diversity and improving recommendation relevance. This work underscores the user intention dimension in the explicit-implicit feedback dichotomy and offers insights for designing personalized recommendation feedback that better responds to users' needs.2025WLWenqi Li et al.Peking University, Department of Information ManagementExplainable AI (XAI)Recommender System UXCHI
EchoSight: Streamlining Bidirectional Virtual-physical Interaction with In-situ Optical TetheringEmerging AR applications require seamless integration of the virtual and physical worlds, which calls for tools that support both passive perception and active manipulation of the environment, enabling bidirectional interaction. We introduce EchoSight, a system for AR glasses that enables efficient look-and-control bidirectional interaction. EchoSight exploits optical wireless communication to instantaneously connect virtual data with its physical counterpart. EchoSight's unique dual-element optical design leverages beam directionality to automatically align the user's focus with target objects, reducing the overhead in both target identification and subsequent communication. This approach streamlines user interaction, reducing cognitive load and enhancing engagement. Our evaluations demonstrate EchoSight's effectiveness for room-scale communication, achieving distances up to 5 m and viewing angles up to 120 degrees. A study with 12 participants confirms EchoSight's improved efficiency and user experience over traditional methods, such as QR Code scanning and voice control, in AR IoT applications.2025JLJingyu Li et al.Peking University, SCSAR Navigation & Context AwarenessContext-Aware ComputingSmart Home Interaction DesignCHI
CrowdBot: An Open-Environment Robot Management System for On-Campus ServicesWang 等人设计 CrowdBot 开放环境机器人管理系统，实现校园场景下机器人的自主导航与任务调度，为校园服务机器人的高效管理提供解决方案。2024YWYufei Wang et al.Domestic RobotsSocial Robot InteractionUbiComp
LoCal: An Automatic Location Attribute Calibration Approach for Large-Scale Deployment of mmWave-based Sensing SystemsZhang 等人提出 LoCal 自动位置校准方法，解决大规模毫米波感知系统部署中的位置属性标定难题，降低系统部署成本与复杂度2024DZDuo Zhang et al.Context-Aware ComputingUbiquitous ComputingUbiComp
Waffle: A Waterproof mmWave-based Human Sensing System inside Bathrooms with Running WaterZhang 等人开发 Waffle 防水毫米波传感系统，专门解决浴室有流水环境中的人体感知难题，实现全天候室内监测。2024XZXusheng Zhang et al.Human Pose & Activity RecognitionContext-Aware ComputingUbiComp
Understanding the Effects of Restraining Finger Coactivation in Mid-Air Typing: from a Neuromechanical PerspectiveTyping in mid-air is often perceived as intuitive yet presents challenges due to finger coactivation, a neuromechanical phenomenon that involves involuntary finger movements stemming from the lack of physical constraints. Previous studies were used to examine and address the impacts of finger coactivation using algorithmic approaches. Alternatively, this paper explores the neuromechanical effects of finger coactivation on mid-air typing, aiming to deepen our understanding and provide valuable insights to improve these interactions. We utilized a wearable device that restrains finger coactivation as a prop to conduct two mid-air studies, including a rapid finger-tapping task and a ten-finger typing task. The results revealed that restraining coactivation not only reduced mispresses, which is a classic coactivated error always considered as harm caused by coactivation. Unexpectedly, the reduction of motor control errors and spelling errors, thinking as non-coactivated errors, also be observed. Additionally, the study evaluated the neural resources involved in motor execution using functional Near Infrared Spectroscopy (fNIRS), which tracked cortical arousal during mid-air typing. The findings demonstrated decreased activation in the primary motor cortex of the left hemisphere when coactivation was restrained, suggesting a diminished motor execution load. This reduction suggests that a portion of neural resources is conserved, which also potentially aligns with perceived lower mental workload and decreased frustration levels.2024HZHechuan Zhang et al.Full-Body Interaction & Embodied InputUIST
Deus Ex Machina and Personas from Large Language Models: Investigating the Composition of AI-Generated Persona DescriptionsLarge language models (LLMs) can generate personas based on prompts that describe the target user group. To understand what kind of personas LLMs generate, we investigate the diversity and bias in 450 LLM-generated personas with the help of internal evaluators (n=4) and subject-matter experts (SMEs) (n=5). The research findings reveal biases in LLM-generated personas, particularly in age, occupation, and pain points, as well as a strong bias towards personas from the United States. Human evaluations demonstrate that LLM persona descriptions were informative, believable, positive, relatable, and not stereotyped. The SMEs rated the personas slightly more stereotypical, less positive, and less relatable than the internal evaluators. The findings suggest that LLMs can generate consistent personas perceived as believable, relatable, and informative while containing relatively low amounts of stereotyping.2024JSJoni Salminen et al.University of VaasaHuman-LLM CollaborationAI Ethics, Fairness & AccountabilityCHI
A Systematic Review of Ability-diverse Collaboration through Ability-based Lens in HCI In a world where diversity is increasingly recognised and celebrated, it is important for HCI to embrace the evolving methods and theories for technologies to reflect the diversity of its users and be ability-centric. Interdependence Theory, an example of this evolution, highlights the interpersonal relationships between humans and technologies and how technologies should be designed to meet shared goals and outcomes for people, regardless of their abilities. This necessitates a contemporary understanding of "ability-diverse collaboration," which motivated this review. In this review, we offer an analysis of 117 papers sourced from the ACM Digital Library spanning the last two decades. We contribute (1) a unified taxonomy and the Ability-Diverse Collaboration Framework, (2) a reflective discussion and mapping of the current design space, and (3) future research opportunities and challenges. Finally, we have released our data and analysis tool to encourage the HCI research community to contribute to this ongoing effort.2024LXLan Xiao et al.University College London, University College LondonCognitive Impairment & Neurodiversity (Autism, ADHD, Dyslexia)Universal & Inclusive DesignInclusive DesignCHI
Make Interaction Situated: Designing User Acceptable Interaction for Situated Visualization in Public EnvironmentsSituated visualization blends data into the real world to fulfill individuals’ contextual information needs. However, interacting with situated visualization in public environments faces challenges posed by users’ acceptance and contextual constraints. To explore appropriate interaction design, we first conduct a formative study to identify users’ needs for data and interaction. Informed by the findings, we summarize appropriate interaction modalities with eye-based, hand-based and spatially-aware object interaction for situated visualization in public environments. Then, through an iterative design process with six users, we explore and implement interactive techniques for activating and analyzing with situated visualization. To assess the effectiveness and acceptance of these interactions, we integrate them into an AR prototype and conduct a within-subjects study in public scenarios using conventional hand-only interactions as the baseline. The results show that participants preferred our prototype over the baseline, attributing their preference to the interactions being more acceptable, flexible, and practical in public.2024QZQian Zhu et al.The Hong Kong University of Science and Technology, The Hong Kong University of Science and TechnologyAR Navigation & Context AwarenessContext-Aware ComputingField StudiesCHI
Exploring the Design of Generative AI in Supporting Music-based Reminiscence for Older AdultsMusic-based reminiscence has the potential to positively impact the psychological well-being of older adults. However, the aging process and physiological changes, such as memory decline and limited verbal communication, may impede the ability of older adults to recall their memories and life experiences. Given the advanced capabilities of generative artificial intelligence (AI) systems, such as generated conversations and images, and their potential to facilitate the reminiscing process, this study aims to explore the design of generative AI to support music-based reminiscence in older adults. This study follows a user-centered design approach incorporating various stages, including detailed interviews with two social workers and two design workshops (involving ten older adults). Our work contributes to an in-depth understanding of older adults’ attitudes toward utilizing generative AI for supporting music-based reminiscence and identifies concrete design considerations for the future design of generative AI to enhance the reminiscence experience of older adults.2024YJYucheng Jin et al.Hong Kong Baptist UniversityGenerative AI (Text, Image, Music, Video)Mental Health Apps & Online Support CommunitiesReproductive & Women's HealthCHI
StarRescue: the Design and Evaluation of A Turn-Taking Collaborative Game for Facilitating Autistic Children's Social SkillsAutism Spectrum Disorder (ASD) presents challenges in social interaction skill development, particularly in turn-taking. Digital interventions offer potential solutions for improving autistic children's social skills but often lack addressing specific collaboration techniques. Therefore, we designed a prototype of a turn-taking collaborative tablet game, StarRescue, which encourages children's distinct collaborative roles and interdependence while progressively enhancing sharing and mutual planning skills. We further conducted a controlled study with 32 autistic children to evaluate StarRescue's usability and potential effectiveness in improving their social skills. Findings indicated that StarRescue has great potential to foster turn-taking skills and social communication skills (e.g., prompting, negotiation, task allocation) within the game and also extend beyond the game. Additionally, we discussed implications for future work, such as including parents as game spectators and understanding autistic children's territory awareness in collaboration. Our study contributes a promising digital intervention for autistic children's turn-taking social skill development via a scaffolding approach and valuable design implications for future research.2024RBRongqi Bei et al.University of MichiganCognitive Impairment & Neurodiversity (Autism, ADHD, Dyslexia)Serious & Functional GamesCHI
LT-Fall: The Design and Implementation of a Life-threatening Fall Detection and Alarming SystemFalls are the leading cause of fatal injuries to elders in modern society, which has motivated researchers to propose various fall detection technologies. We observe that most of the existing fall detection solutions are diverging from the purpose of fall detection: timely alarming the family members, medical staff or first responders to save the life of the human with severe injury caused by fall. Instead, they focus on detecting the behavior of human falls, which does not necessarily mean a human is in real danger. The real critical situation is when a human cannot get up without assistance and is thus lying on the ground after the fall because of losing consciousness or becoming incapacitated due to severe injury. In this paper, we define a life-threatening fall as a behavior that involves a falling down followed by a long-lie of humans on the ground, and for the first time point out that a fall detection system should focus on detecting life-threatening falls instead of detecting any random falls. Accordingly, we design and implement LT-Fall, a mmWave-based life-threatening fall detection and alarming system. LT-Fall detects and reports both fall and fall-like behaviors in the first stage and then identifies life-threatening falls by continuously monitoring the human status after fall in the second stage. We propose a joint spatio-temporal localization technique to detect and locate the micro-motions of the human, which solves the challenge of mmWave's insufficient spatial resolution when the human is static, i.e., lying on the ground. Extensive evaluation on 15 volunteers demonstrates that compared to the state-of-the-art work (92% precision and 94% recall), LT-Fall achieves zero false alarms as well as a precision of 100% and a recall of 98.8%. https://dl.acm.org/doi/10.1145/35808352023DZDuo Zhang et al.Elderly Care & Dementia SupportBiosensors & Physiological MonitoringUbiComp
WiMeasure: Millimeter-level Object Size Measurement with Commodity WiFi DevicesIn the past few years, a large range of wireless signals such as WiFi, RFID, UWB and Millimeter Wave were utilized for sensing purposes. Among these wireless sensing modalities, WiFi sensing attracts a lot of attention owing to the pervasiveness of WiFi infrastructure in our surrounding environments. While WiFi sensing has achieved a great success in capturing the target's motion information ranging from coarse-grained activities and gestures to fine-grained vital signs, it still has difficulties in precisely obtaining the target size owing to the low frequency and small bandwidth of WiFi signals. Even Millimeter Wave radar can only achieve a very coarse-grained size measurement. High precision object size sensing requires using RF signals in the extremely high-frequency band (e.g., Terahertz band). In this paper, we utilize low-frequency WiFi signals to achieve accurate object size measurement without requiring any learning or training. The key insight is that when an object moves between a pair of WiFi transceivers, the WiFi CSI variations contain singular points (i.e., singularities) and we observe an exciting opportunity of employing the number of singularities to measure the object size. In this work, we model the relationship between the object size and the number of singularities when an object moves near the LoS path, which lays the theoretical foundation for the proposed system to work. By addressing multiple challenges, for the first time, we make WiFi-based object size measurement work on commodity WiFi cards and achieve a surprisingly low median error of 2.6 mm. We believe this work is an important missing piece of WiFi sensing and opens the door to size measurement using low-cost low-frequency RF signals. https://dl.acm.org/doi/10.1145/35962502023XWXuanzhi Wang et al.Context-Aware ComputingUbiquitous ComputingComputational Methods in HCIUbiComp
BLEselect: Gestural IoT Device Selection via Bluetooth Angle of Arrival Estimation from Smart Glasses"Spontaneous selection of IoT devices from the head-mounted device is key for user-centered pervasive interaction. BLEselect enables users to select an unmodified Bluetooth 5.1 compatible IoT device by nodding at, pointing at, or drawing a circle in the air around it. We designed a compact antenna array that fits on a pair of smart glasses to estimate the Angle of Arrival (AoA) of IoT and wrist-worn devices' advertising signals. We then developed a sensing pipeline that supports all three selection gestures with lightweight machine learning models, which are trained in real-time for both hand gestures. Extensive characterizations and evaluations show that our system is accurate, natural, low-power, and privacy-preserving. Despite the small effective size of the antenna array, our system achieves a higher than 90% selection accuracy within a 3 meters distance in front of the user. In a user study that mimics real-life usage cases, the overall selection accuracy is 96.7% for a diverse set of 22 participants in terms of age, technology savviness, and body structures. https://dl.acm.org/doi/10.1145/3569482"2023TZTengxiang Zhang et al.On-Skin Display & On-Skin InputContext-Aware ComputingUbiquitous ComputingUbiComp
Understanding Disclosure and Support in Social Music Communities for Youth Mental HealthOnline music platforms that embed social features are enabling the creation of supportive social communities where many young people disclose their distressing feelings and seek support. However, there is a limited understanding of the content young people disclose or the support they may provide in such social music communities. In this work, using a large online music platform as our research site, we employed mixed methods to analyze users' comments (N=163) and the associated replies (N=2,732) related to young people's psychological distress (e.g., depression, anxiety, stress, and loneliness). We found that experience sharing dominates the types of comments, which often invokes peers' support in the form of encouragement, caring, or self-disclosure. Furthermore, we conducted an interview study with 13 young people to understand their perceptions of and motives for disclosure and support on our research site. The interviewees expressed that music-induced and comment-induced emotional resonance is the main drive for their disclosure and support. Finally, we discuss design implications for a supportive social music community that could benefit youth mental health.2023YJYucheng Jin et al.Mental Health IICSCW