Kai Yang      

Currently, I am a second-year master student at University of Electronic Science and Technology of China(UESTC), supervised by Prof. Fan Zhou and Prof. Ting Zhong. In addition, I have been an IEEE student Member since 2023. My research interests include but are not limited to Trustworthy AI, Spatial-temporal Recommendation System.

Email: kaiyang.cs AT outlook.com  /  Google Scholar  /  Github

We introduce a novel framework called ExNext for Point of Interest (POI) recommendation in LBSNs, which focuses on improving the trustworthiness of recommender systems by enhancing both accuracy and explainability. Our framework utilizes information theory to learn a compact representation that balances knowledge compression with retaining relevant information.

We endow the geolocation model with explainability through variational IB theory, which aims at striking a balance between knowledge compression (provid-ing concise explanations) and preserving pertinent information (offering informative representations).

We provide a new view of geolocation through mitigating sparsity, which aims at enhancing contextual information utilization in social graph and network topology.

We provide an innovative method of event prediction, which not only exhibits remarkable precision in entity predictions but also enhances interpretability by identifying pivotal historical events influencing event predictions.