Information extraction (IE) has been widely studied in various domains since IE is a key to bridge the gap between knowledge and texts. IE includes several core sub-problems, such as named entity recognition, relation extraction, and event extraction, and these sub-problems have been tackled using machine learning techniques. In this talk, I will give an overview of learning approaches for IE in biomedical and general domain, especially on corpus-based classification and structured learning approaches. I will then introduce recent deep learning approaches including our recent recurrent neural network (RNN)- based approach, and discuss the limitations and future directions.
Makoto Miwa is an associate professor of Toyota Technological Institute (TTI). He received his Ph.D. from the University of Tokyo in 2008. His research mainly focuses on information extraction from texts, deep learning, and representation learning. His projects include AkaneRE, EventMine, PathText and LSTM-ER.