This line of research started with my training in graduate school and explores how human relationships evolve in the context of computer-mediated communication and how they may differ from its counterpart offline.
Awareness information checking and its influencing factors
Informal interaction online is commonly supported by providing awareness information of each other’s presence and activities, and providing such information can help one time one's initiation of interaction strategically.
In this project, we looked at the factors that may affect those checking behaviors, such as the perceived urgency of one’s motivation for checking, the visibility of these checks, and one’s cultural background.
We found that awareness checks occurred more frequently when they were invisible to the interactants' partner and when the task at hand was constrained in time. In a follow-up experiment, we also found that members of collectivistic cultures adjusted their behaviors to match those of their partners, whereas members of individualistic cultures did not display such accommodation.
Availability management and deception
The omni-connectedness of current information and communication technologies (ICTs) enables us to reach others and be reachable almost whenever and wherever we are. Still, it also brings inconveniences when one does not want to be contacted. People have developed strategies to manage their availability in situations like this.
One of the most used strategies is butler lies. This project uses online surveys to explore how the availability management strategies, primarily butler lies, become the social lubricant in different interpersonal relationships (e.g., friends, colleagues, and family members) in a U.S. sample.
We found that butler lies occurred the most frequently with acquaintances and groups of acquaintances, and the least with significant others. We also found such behaviors were moderated by people’s conversational grounding levels and trust for each other.
User experience studies in localized contexts
This line of research centers around the user experience concept and explores the adoption and adaptation of technologies by different users to achieve their goals.
Technology addiction and abuse in the adolescent population
Smartphones and other relevant technologies are often considered the culprit of technology abuse and setting limits for screen time is one primary strategy to address it, especially for adolescents who are still living with their parents.
However, such externally imposed temporal and behavioral structure often incurs conflicts among family members and does not often realize the goal it sets out to accomplish. This project is interested in understanding the data-sharing behaviors between adolescents and their parents as a measure to combat technology addiction and abuse.
We found that involving parents in the treatment process of technology abuse may be an effective strategy if data are properly shared between parents and children.
Social sensing and gig-workers
The popularity of the gig economy seems to offer the workers more flexibility in scheduling their work time and adjusting the workload. Still, due to the low rate, many workers are used to extending their work hours to trade for a higher payment, leading to health risks.
In this project, we looked at how wearable technologies can be integrated in the gig worker's social network to promote their work-life balance.
We found that while technological sensing was able to increase drivers’ awareness of their well-being status and intention to reduce over-working, social sensing goes a step further and increased awareness for all the family members involved (in our cases, the drivers and their spouses) and was able to change the actual behaviors of the drivers rather than just intentions.
User-centered artificial intelligence design
This line of research revolves around the technological designs relating to artificial intelligence and examines how users react to design elements in prototypes.
Multimodal sensing and task-oriented chatbot
Chatbots can be dichotomized into task-oriented and social-oriented types. In the context of using a task-oriented chatbot, users often need to describe their immediate need and contextual information relevant to such a need so that the chatbot can pinpoint what help to offer, which is usually a challenging task for users.
This is an interdisciplinary project, where we designed a chatbot that helped the users complete an assembly task that simulated those in manufacturing settings. By classifying the users’ intents, the chatbot was able to provide answers or respond with proper instructions to the user when he/she encountered problems during assembly.
We pushed the project further and proposed to incorporate the visual cues to increase intent classification accuracy.
Hateful memes and explainable AI
Memes are popular rhetorical devices on social media and sometimes disseminate hatred, racism, and extreme political views. Although many platforms take active measures to remove such content from the audience, the current detection mechanism still falls behind the need.
We argue that as a part of the explainable AI (XAI) endeavor, human-comprehensible explanations should accompany the memes to raise users’ awareness of hateful memes and, in a way, immunize them for future encounters.
We found that people without prior cultural knowledge of the hateful meme gained significantly more understanding and awareness of the meme when presented with the complex explanation generated and edited by crowd workers than in the other two conditions. In addition, although the explanation was more complex and might take longer to read, it did not impose any extra cognitive workload for the users.