top of page
AI for Cognitive Reappraisal
Project | Designing an AI system to Facilitate Cognitive Reappraisal from Peer Support
My Role:  I supervise and manage this project
Method: Mixed-Methods Research, Deep learning
Time: 2021 - present

Reappraisal is an important technique for people to change their negative thoughts and emotions, and this approach usually involves reinterpreting and distancing the distress events. Yet, it is not always easy for everyone to apply this method flexibly, so people may turn to online support communities to seek support and advice. However, how support providers could contribute to support seekers' reappraisal is under-explored. This work deployed an online support community and conducted a field study to collect support providers' comments on helping support seekers' reappraisal. We then worked with licensed psychotherapists to examine how support providers generated messages to support the reappraisal process. Our results demonstrate the feasibility of using AI technology to coordinate support providers' comments to facilitate reappraisal. We finally discuss design considerations and implications that could help the future design of the online support community to promote users' mutual help for elicit positive reappraisal.

Chatbots for Social Learning
Project | Incorporating Social Learning to Enhance Chatbot-guided Learning Experience
My Role: Project Lead
Method: Mixed-Methods Research, Chatbot system
Time: 2020 - present

Chatbots are promising for providing guidance for users to practice various skills; however, the challenge of engaging people to put effort into the practice still remains. The present study addresses this challenge by incorporating a social-learning component(i.e., learning from peers' experiences) into human-chatbot interaction. We designed two chatbots that deliver guidance for the practice of journaling skills, one with and one without a social-learning component. Our three-week user study showed that the participants talking to the social-learning chatbot put significantly more effort into practicing the journaling skills; although they were not able to have dyadic interactions with their peers, many of them delegated the chatbot to share positive experiences of their practices and suggestions with others; additionally, some participants' perceptions of the practices were biased when the chatbot gate-kept the social-learning process by only sharing peers' positive experiences. These findings provide design implications for delivering effective skills-training interventions via chatbots.

LGBTQ+ Relationship Building
Project | Exploring LGBTQ+ Relationship Building via Online Dating
My Role: I supervise and manage this project
Method: Qualitative-method Research, Interview
Time: 2021 - present

In some Asian countries, lesbian, bisexual, and transgender women – sexual-minority women (SMWs) – are strongly stigmatized and have limited opportunities to connect with other SMWs in offline contexts. Although dating apps help them connect and seek social support, little is known about SMWs’ practices of self-disclosure and connection-building through those apps. Therefore, we interviewed 43 SMW dating-app users in China to explore their experiences of utilizing such apps to build relationships. We found that these apps were a double-edged sword: providing a space in which SMWs could disclose their true selves while heightening their fears of becoming targets of discrimination if their information leaked out. To minimize such risks, they developed distinctive strategies for disclosing themselves. Additionally, they shared how they used dating apps to recognize other SMWs offline and build relationships with them. Our findings have design implications for supporting SMWs and improving their online dating experiences.

Project | Exploring the Effect of Chatbots’ Language Style on Users’ Learning Outcomes
My Role: I supervise and manage this project
Method: Mixed-methods Research
Time: 2020 - present

Chatbots can assist learners by providing a relaxing environment and opportunities for conversational practice. However, it is not yet known whether chatbots’ language styles have a contextual influence on learning outcomes. Therefore, we designed four chatbots, two that spoke to users in a formal style and two that spoke to them in an informal style, using stories to introduce new formal(traditional idioms) or informal(slang) phrases. We then conducted an experiment to examine each chatbot regarding learning phrases over seven days. Our findings reveal that when conversing the chatbots’ language style matched the context of the phrases, users would try to respond in a similar style. Furthermore, we found that language styles provide contextual influence for learning outcomes. When the chatbot’s language style did not match the phrase’s context, its users’ learning outcomes were impeded. Finally, we discuss the implications of our findings for the future design of chatbot-assisted learning.

Chatbot Language Style
Project 1
Project | 01

Publications: CHI2020, CSCW2020

Special Thanks NTT sponsored my thesis research

doctor model.png
Ph.D. Thesis Research | Designing Conversational Agents to Promote Self-disclosure and Behavior Change 
Role: Project Lead

     Conversational agents (CAs), as an AI technology, are regarded as one of the most promising technologies and are increasingly applied in many domains. Because CAs provide a fast, convenient, and low-cost communication channel, both scholars and practitioners are keen to develop effective CAs to address the challenges of providing healthcare services and improving people's well-being.

However, effective communication with people through CAs requires people's trust and people's willingness to self-disclose their personal information or experiences with the CAs. To achieve long-term benefits (e.g., addressing chronological health issues), being able to sustain communication between people and the CAs also remains a challenge.

In this dissertation research, we design, implement and evaluate CAs to address the fundamental challenges of effectively eliciting and sustaining people's self-disclosure with the CAs in different social contexts. More specifically, our research will contribute to:  1) effective CA designs that succeed in eliciting people's deep self-disclosure to a CA over time, 2) empirical evidence and deeper understandings of sustaining people's trust of CAs throughout different interaction periods, e.g., with and without the involvement of real mental health professionals, and  3) new design insights of integrating human support in human-CA interaction to promote behavior change. The global pandemic (COVID-19) heightens the challenges of providing healthcare services at all levels, our research will make a timely impact on society and a lasting impact on human-AI research at large.  

Project | 02
Independent Project | Investigating Social Interaction and Tipping Behaviors in Live Streaming
Live streaming (e.g., Twitch, Youtube Live) is a proliferating social media generating special social interactions on the platforms. For example, tipping in live streaming, in which viewers buy virtual gifts to reward the streamers. This project explores people's tipping behaviors and how it impacts interactions between viewers and streamers. We analyzed videos of live streams by labeling the viewer-streamer interactions and found that viewers were motivated by the reciprocal acts of streamers. Furthermore, we interviewed many streamers to understand their motivations behind the viewer-streamer interactions. This research contributes to the understanding of tipping behaviors from both streamers and viewers' perspectives. We have published some research papers, please refer them for more detail.
Project 2

Publications: CHI2018, CSCW2019, MobileHCI2019

Project | A Smart Crowd Donation Platform for Supporting Education and Classroom

Donation-based crowdfunding has the potential to democratize capital raising by soliciting donations directly from the public through the Web and social media. These crowdfunding platforms, however, often function as unregulated open markets, in which there is a minimal intervention to influence donation distribution across projects. In fact, research on crowdfunding hints that donation distribution in most crowdfunding platforms is suboptimal: while the overall success rates of crowdfunding projects are often low, a significant proportion of projects receive donations way over their targets. In this paper, we propose a new donation distributing system that aims to (a) distribute donations more effectively among the projects, and (b) align the allocation of donations with the preferences of donors. An agent-based model was developed to test the proposed system. Results showed that the proposed system not only increased the overall success rates of projects but also led to more successes for projects preferred by donors. Implications to future crowdfunding platforms are discussed.

Project | 03

Publications: IUI 2018, SBP2017, CHI2015, SBP2016

Project 3
Screen Shot 2020-07-10 at 10.23.37
Project 4
Project | 04
Project | Using Time-Anchored Peer Comments to Enhance Social Interaction in Online Educational Videos


Online learning is increasingly prevalent as an option for self-learning and as a resource for instructional design. Prerecorded video is currently the main medium of online education content delivery and instruction; this affords asynchronicity and flexibility and enables the dissemination of lecture content in a distributed and scalable manner. However, the same properties may impede learners' engagement due to the lack of social interaction and peer support. In this paper, we propose a time-anchored commenting interface to allow online learners who watch the same video clips to exchange comments on them. Comments left by previous learners at specific time points of a video are displayed to new learners when they watch the same video and reach those time points. We investigated how the display of time-anchored comments (dynamic or static) and type of comments (content-related or social-oriented) influenced users' perceived engagement, perceived social interactivity, and learning outcomes. Our results show that dynamically displaying time-anchored comments can indeed enhance learners' perceived social interactivity. Moreover, the content of comments would further affect learners' intention of commenting. Based on our findings, we make various recommendations for the improvement of social interaction and learning experience in online education.

Publications: CHI 2015

Screen Shot 2020-07-10 at 9.59.14 AM.png
Project 5
Project | 05

Publications: Sensors 2020, TNSRE 2016, CHI 2014, TIM 2015

Screen Shot 2020-07-02 at 2.27.27 PM.png
Master Degree Thesis Research | Applying EEG Analysis to Human-Computer Interaction Design

     Human cognitive limitations are becoming a crucial issue in human-computer interaction nowadays and thus an important determinant of overall human performance. This thesis purpose to introduce two methods that applying EEG technology helps us to improve and evaluate human-computer interaction.

We suggested a new approach to detect users’ attention state. Attention monitoring is particularly important for many HCI applications. How to automatically determine a user's visual attention state is challenging since attention involves many complex and internal human cognitive functions. Behavioral observations, such as eye gaze or response to external stimuli, can provide some clues for users' visual attention state; however, the users' cognitive state cannot be easily known. We explored the feasibility of designing an attention monitoring system that can detect if our brain sees a visual stimulus consciously.

Moreover, we used an EEG-based approach to assist usability tests with audio notifications. Audio notifications have become an important way to prompt users. Several studies have been proposed to evaluate audio notifications, but they rarely considered user workload and environmental impact at the same time. We developed an EEG-based approach to evaluate audio notifications by measuring subjects’ auditory perceptual response (mismatch negativity) and attention status (P3a). We demonstrated this approach by two experiments, in which auditory icons were evaluated under different workloads and environments. According to the experiment results, the perceptual effects of audio notifications could be measured objectively.


Based on these technologies, we could directly detect users’ cognitive changes when they interact with our design. Therefore, this thesis used an EEG device to detect and analyze the variation of human cognitive state from visual and auditory perception respectively. We hope to provide new feasible approaches and views to the HCI field.

Screen Shot 2020-07-10 at 10.07.29
bottom of page