【Ming Li Lecture Hall, 2020 Issue 5】Professor Wei Zhiping, Department of Information Management, National Taiwan University: Natural Language Understanding of Biomedical Literature: Biomedical Relation Extraction Methods and Their Applications

Time: January 7th (Thursday) 14:30 PM-16:00 PM

Tencent meeting number: 296 459 659

Report Content Description:

The size of biomedical literature is massive and expands at a fast rate, due to the rapid growth in biomedical research and development. PubMed is an online portal (accessing primarily the MEDLINE database) that include more than 30 million of research articles (abstracts) on life sciences and biomedical topics by the end of January 2020. Biomedical literature provides healthcare practitioners (e.g., physicians, pharmacists) up-to-date biomedical research findings, which can be applied to improve professional practices and healthcare outcomes. Moreover, biomedical literature is core to new knowledge creation and discovery. Because the size of biomedical literature expands rapidly, manual review and inspection of biomedical research articles is very difficult and time-consuming. As a result, the development of some natural language understanding (NLU)techniques that can comprehend or extract important information from this huge collection of literature is essential and desirable.

One important type of information that can be extracted from these articles are biomedical relations discussed in each article. Examples of biomedical relations include drug-disease relations, chemical-protein relations, gene-disease relations, protein interactions, drug-drug interactions, etc. Formally, given a sentence (or a small segment of text) that contains two entities of interest, the task of relation extraction is to predict whether the sentence describes some relation (out of a predefined set of relation types) between the two entities and, if so, to classify which relation class does the sentence point to. In this talk, I will present our proposed biomedical relation extraction methods that follow the deep-learning-based approach. In addition, in this talk, I will also discuss an important application of biomedical relation extraction, i.e., literature-based drug repurposing.

Speaker profile:

Professor Wei Zhiping currently works as a distinguished professor in the Information Management Department of National Taiwan University. Professor Wei holds a Ph.D. in Management Information System from University of Arizona (graduated in 1996). He has taught at Tsinghua University and Sun Yat-sen University. He has also served as a visiting scholar at the University of Washington, the University of Illinois at Urbana-Champaign, and the Chinese University of Hong Kong.

Professor Wei's main research fields are big data analysis, text mining, social media analysis, biomedical information, patent analysis and exploration, etc. His research results are published in internationally renowned journals in information management or information technology related fields, such as Journal of Management Information Systems (JMIS), European Journal of Information Systems (EJIS), Decision Sciences, Decision Support Systems (DSS), Information & Management (I&M), Journal of the Association for Information Science and Technology, IEEE Transactions in Engineering Management, IEEE Transactions on Systems, Man, and Cybernetics, IEEE Intelligent Systems, IEEE Transactions on Information Technology in Biomedicine, etc.

(Organized by: Department of Management Engineering, Scientific Research and Academic Exchange Center)