Department of Advanced Information Technology, Faculty of Information Science and Electrical Engineering
Department of Information Science and Technology, Graduate School of Information Science and Electrical Engineering
Department of Electrical Engineering and Computer Science, School of Engineering
We aim to study human-centered intelligence. To this end, we analyze real data under real situations and develop mechanisms to estimate, extract, and generate information users want and provide it to them when they need, considering their contexts, intentions, preferences, interests, and privacy issues. The projects we are conducting are roughly divided into four: 1) Text Mining and Message Generation, 2) Data Mining for Intelligent Transport Systems (ICT), 3) Educational Data Mining (EDM), and 4) Multi-modal Data Mining and Information Recommendation. For 1), we develop dialogue systems (Chat-bots) which automatically answer user queries, discriminating out-of-domain or out-of-intent queries with query augmentation techniques; we estimate user emotions; we study named entity recognition from patent documents and research papers, etc. For 2), we estimate city bus travel time, arrival time, and delay time, abnormal driving behaviors, and road situations by analyzing multi-modal ICT-related data such as vehicle probe data, obtained from ICT-devices (ETC 2.0 devices), dashboard camera data, weather-related data, traffic and human stream data, etc. For 3), we develop methods to estimate student learning situations and performance, to give automatic feedback analyzing student data such as student self-reflective comments freely-written after each lesson, e-learning logs etc., and to automatically score student short answers. Finally, for 4), we estimate recommended handcrafted works, which work will be bought in certain period of time, and who created the works, and track trends or changes of the works; we also estimate useful product review documents, and develop new collaborative filtering algorithms using Graph Convolutional Networks to extract useful information from user-item interactions.
Assoc.Prof. Tsunenori Mine