Junjie Ma's research sits within Human-Computer Interaction (HCI), with a focus on Human-Robot Interaction (HRI) and Human-AI Interaction. The central direction is the design of personalized narrative mechanisms for Embodied AI, physical or socially present systems that interact with people in real-world settings. The core argument is that Embodied AI should carry highly customized, differentiated narrative modes built around what humans actually need and expect.
This work unfolds across two dimensions. The first, Narrative Style, concerns personalization settings, persona design, and expressive behavior, an area that draws on psychology, sociology, and cognitive science. The second, Narrative Themes, concerns cultural sensitivity and multilingual ability within social contexts. Together, these dimensions converge on three focal areas: Expression Mechanisms, Cultural AI, and Multilingualism.
Expression Mechanisms
Research on expression mechanisms for Embodied AI spans many directions. Junjie Ma's focus is on building behavioral and emotional expression systems through minimal technical intervention, without software iteration overhead, add-on hardware, or changes to physical form. The goal is to allow Embodied AI to work with communication patterns already established in human social life, so that norms developed in real or online communities can carry over directly into human-robot and human-AI interaction. This lowers the learning cost and cognitive load on users, letting them stay focused on their core tasks.
Cultural AI
Across different cultural contexts, Embodied AI must be sensitive and adaptable enough to behave appropriately with people from varied ethnic and cultural backgrounds. Junjie Ma's work in this area centers on two mechanisms: how user communities can configure and customize Embodied AI behavior; and how Embodied AI can learn and apply cross-cultural behavioral norms through contextual awareness and ongoing interaction.
Multilingualism
In globally deployed and cross-cultural contexts, multilingual capability is a key factor in how broadly useful Embodied AI can be. Junjie Ma's research addresses the barriers that arise in cross-language narration, aiming to maintain semantic equivalence and minimize information loss across languages. The focus is not only on accurate language conversion, but on keeping narrative style consistent and culturally appropriate across different language environments, so that interaction remains smooth and coherent.
Broader HCI Interests
Beyond the core focus on Embodied AI, Junjie Ma has engaged with a wider range of projects across Human-Computer Interaction. These projects do not fall neatly under the themes of Expression Mechanisms, Cultural AI, or Multilingualism, yet each reflects a genuine curiosity about how people interact with technology and with one another through it. Topics explored include peer production and collaborative knowledge construction, algorithmic and model efficiency in sociotechnical systems, data engineering practices, and bias in data and computational models. Taken together, they reflect an interest in HCI as a broad and pluralistic discipline — one in which diverse problems, methods, and communities of practice continually inform one another.
Advisors and Affiliations
| Role | Person / Institution |
|---|---|
| Ph.D. Advisor | Dr. Zhicong Lu, DEER Lab, George Mason University |
| M.S. / B.S. Advisors | Dr. Loren Terveen and Dr. Stevie Chancellor, University of Minnesota |
See Also
- Publications — Papers and projects arising from this work
- Teaching — Courses related to these themes
- Home — Back to main page