Georgia Institute of Technology
Ruohao’s Ph.D. research focuses on designing conversational AI systems that provide instructions for daily tasks such as cooking, specifically catering to older adults diagnosed with MCI and their care teams. The primary goal of her research is to develop a user-friendly conversational AI system for cooking instructions that adapts to users’ preferences and feedback, regardless of their level of kitchen experience. This system will be designed to generate customized cooking instructions that take into account users’ preferences, dietary restrictions, and ingredient availability while ensuring that the recipe steps are followed accurately. To achieve this goal, Ruohao collected a dataset of 270 conversations between users and cooking experts. The dataset was annotated with essential information such as ingredients, recipe steps and dialogue intentions that are necessary for developing accurate and flexible conversational AI systems. By using natural language processing and machine learning algorithms, this dataset has been used to train and develop language models that can generate accurate and personalized cooking instructions during conversations with people.
Moreover, Ruohao is working on textual style understanding, an important area of research in natural language processing with many potential applications, such as sentiment analysis and content recommendation. Understanding the styles behind text can be challenging, especially when working with limited resources. To address this problem, Ruohao proposed a method for style prediction with words that express or describe the style. The method involves teaching a language model to predict styles with instructions and style lexicon information. The results suggest that this approach is effective in predicting new styles based on their lexicon only, without the need of creating a new style dataset. Ruohao hopes that their work will inspire further research in the area of textual style understanding, resulting in the advancement of models capable of providing better personalized assistance to older individuals.
To summarize, Ruohao’s research has focused on developing user-friendly conversational AI systems that generate personalized cooking instructions for older adults with MCI and their caregivers. Her work also includes improving models for textual style understanding, which could have various applications in providing better personalized assistance. Overall, her research has contributed to advancing the development of AI tools that assist and improve the lives of older individuals.