Yankuan Liu

Yankuan Liu

Auto-generated profile, based on publicly available information

Yankuan Liu

Tsinghua University

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
23 September 2025 in Information Technology & People
Verified
Journal Article
Closed access
YLYankuan Liu
Yankuan Liu
PRPei-Luen Patrick Rau
Pei-Luen Patrick Rau
Purpose: This study develops a multi-level framework to examine user perceptions of large language models (LLMs). By investigating the relationships among scenario characteristics (irreplaceability, audience and purpose), user perceptions (familiarity, trust and attitudes) and usage contexts (individual, organizational and societal), this study explores how users evaluate LLM-based chatbots in various contexts, addressing a gap in existing research on LLM applications and their user acceptance. Design/methodology/approach: Utilizing a mixed-methods approach, this study crafted eight diverse usage scenarios spanning niche entertainment to significant production support. A questionnaire survey captured users’ familiarity, trust and attitudes, enabling a systematic analysis of user perceptions at individual, organizational and societal levels. Hierarchical clustering and linear regression linked user perceptions with scenario characteristics, forming the basis of the proposed multi-level framework. Findings: Through the developed framework, this study identifies distinct perception patterns that systematically align with individual, organizational and societal levels of application. For instance, personal assistance scenarios receive higher perception scores, while societal-level applications reveal mixed user perceptions. These insights suggest that enhancing system performance and addressing user concerns are crucial for effective adoption. Originality/value: This study introduces a user-centric, multi-level framework for analyzing LLM perceptions across contexts. It highlights the critical role of user perspectives in shaping ethical, reliable and widely accepted LLM applications, offering practical guidance for developers, organizations and policymakers.

28 July 2025 in JMIR Formative Research
Verified
Journal Article
Open access
XLXiaojun Lai
Xiaojun Lai
LQLiyan Qiao
Liyan Qiao
PRPei-Luen Patrick Rau
Pei-Luen Patrick Rau
YLYankuan Liu
Yankuan Liu
Background: Cerebral small vessel disease (CSVD) significantly impacts motor functions, particularly gait dynamics. However, its analysis often lacks the integration of comprehensive tools that capture the multifaceted nature of gait disturbances. Traditional methods may not fully address the complexity of CSVD’s impact on gait, underscoring the need for a detailed exploration of gait characteristics through advanced technological means. Objective: This study aims to identify the distinct gait patterns and postural adaptations present in patients with CSVD compared to a healthy older population, using an integrative analysis combining sensor and video data to provide a holistic understanding of gait dynamics in CSVD. Methods: This study involved 90 participants older than 50 years (mean age 68.85, SD 9.74 years; 47 males and 43 females), with 24 categorized as normal controls (mean age 66.42, SD 7.51 years) and 66 diagnosed with CSVD (mean age 69.74, SD 10.37 years). Participants performed three walking tasks: normal walking, dual-task walking (with concurrent mental arithmetic), and fast walking. Gait parameters were collected through video data for image posture parameters using the OpenPose BODY_25 key point model, and the “Pocket Gait Test” smartphone app for sensor-based parameters sampled at approximately 40 Hz. Data analysis included 5 sensor-based parameters (step frequency, root mean square (RMS), step variability, step regularity, and step symmetry) and 6 key video-based parameters (including knee angle, ankle angle, elbow angle, body trunk angles, and head posture). Results: Among the 29 participants with complete sensor and video data (10 normal controls and 19 patients with CSVD), significant differences were observed in step regularity (normal walking: mean 0.76, SD 0.09 vs mean 0.61, SD 0.25; P<.003 and dual-task: mean 0.74, SD 0.13 vs mean 0.57 SD 0.24; P<.005), RMS (normal walking: mean 1.64, SD 0.45 vs mean 1.43, SD 0.42; P<.006), and forward head posture angles (head-to-body angle during normal walking: mean 132.96, SD 7.78 vs mean 128.07, SD 7.99; P<.02 and head-to-ground angle: mean 134.11, SD 8.28 vs mean 128.40, SD 9.75; P<.008) between the CSVD and control groups. The CSVD group exhibited a more pronounced forward head posture across all walking tasks, with the greatest difference observed during dual-task walking (head-to-ground angle: mean 134.43, SD 8.29 vs mean 125.02, SD 8.42; P<.02). Conclusions: The study provides compelling evidence of distinct gait disturbances in patients with CSVD, characterized by reduced step regularity (15%‐23% lower than controls), altered acceleration patterns, and significant postural adaptations, particularly forward head positioning (4°‐7° more pronounced than controls). These quantifiable differences, detectable through accessible smartphone and video technology, offer potential biomarkers for early CSVD detection and monitoring. The integration of sensor and video analysis provides a more comprehensive assessment approach that could be implemented in both clinical and home settings for longitudinal monitoring of disease progression and rehabilitation outcomes.

1 June 2024
Verified
Book Chapter
Closed access
PDF file
JZJingyu Zhao
Jingyu Zhao
PRPei-Luen Patrick Rau
Pei-Luen Patrick Rau
YLYankuan Liu
Yankuan Liu



15 June 2026 in ACM Transactions on Human-Robot Interaction
Verified
Journal Article
Open access
YLYankuan Liu
Yankuan Liu
PRPei-Luen Patrick Rau
Pei-Luen Patrick Rau
As robots increasingly collaborate with humans in dynamic environments, the tension between communication clarity and social appropriateness presents an important design challenge. While communicative clarity is traditionally assumed to enhance human–robot interaction, recent findings suggest that clear communication can backfire when robots violate social norms. This study introduces and validates a “social amplifier” model of embodied communication, proposing that the value of role clarity depends on the appropriateness of the robot’s social behavior. We developed a novel confidence-aware role adaptation paradigm where an NAO robot dynamically switches between socially appropriate and inappropriate roles based on users’ confidence levels in an ambiguous video discrimination task. Using a within-subjects design with 64 participants, we manipulated role complementarity, choice alignment, and embodied communication across 512 total interactions. Results revealed that embodied communication significantly enhanced role clarity but had conditional effects on collaborative outcomes. Contrary to the “bad is stronger than good” principle, embodied communication asymmetrically amplified positive effects of appropriate robot behavior while having minimal impact on inappropriate behavior. This pattern emerged consistently across subjective measures (collaboration quality, role dynamics, trust) and objective behavioral outcomes (confidence change, choice change). The findings also demonstrate that robots with socially appropriate roles became over 7 times more persuasive when enhanced with rich embodied cues, while inappropriate robots showed no benefit. These results provide evidence-based design principles prioritizing social intelligence over mere expressiveness in collaborative decision-making robots.