Workshop

Date: 15.05.2015 (Friday)
Time: 14:00
Place: B-103 (Beseri Bilimler Binasi)

First Session:
Title: Knowledge Representation and Reasoning for Drug Discovery

Presented By Esra Erdem (Sabanci University)

Abstract. We present new methods to efficiently answer complex queries over biomedical ontologies and databases considering the relevant parts of these knowledge resources, and to generate shortest explanations to justify these answers. Both algorithms rely on the high-level knowledge representation formalism and efficient solvers of Answer Set Programming. We apply these algorithms to find answers and explanations to some complex queries related to drug discovery, over PHARMGKB, DRUGBANK, BIOGRID, CTD and SIDER.

Short- Bio: Esra Erdem is an associate professor in Computer Science and Engineering at Sabanci University. She received her Ph.D. in computer sciences at the University of Texas at Austin (2002), and visited University of Toronto and Vienna University of Technology for postdoctoral research (2002-2006). Her research is in the area of artificial intelligence, in particular, the mathematical foundations of knowledge representation and reasoning, and their applications to cognitive robotics and computational biology.


Second Session:
Title: Linked Data as a Reference Knowledge Resource for Question Answering and Information Extraction

Presented By Erdogan Dogdu (TOBB EDU)

Abstract. Semantic Web is the next chapter of the Web and Linked Data has been a practical manifestation of structured semantic Web with huge amounts of data collected from crowd sourced information resources, specifically Wikipedia and the like. In this talk we will discuss the importance of linked data as a reference semantic knowledge resource for many computational tasks, specifically information extraction and question answering. Recent intelligent software systems and machine learning algorithms started to utilize linked data to accomplish human-like tasks such answering natural language questions.

In the first part of the talk, we will talk about our work on how we utilize linked data to answer natural language questions. And, in the second part of the talk we will talk about another usage of linked data in document classification. Earlier work in document classification utilized word frequencies such TF-IDF method. We utilize linked data to better identify named entities and therefore classify documents much more accurately. We tested our methods on 20-newsgroup dataset. The results are promising and have shown 40% more accuracy compared to TF-IDF method.

Short-Bio: Erdogan Dogdu received his MS and PhD in Computer Science from Case Western Reserve University (Cleveland, USA) in 1992 and 1998 respectively and BS degree in Computer Science and Engineering from Hacettepe University (Ankara) in 1987. He worked as a project manager and software engineer at VoiceCom (Cleveland and Atlanta, USA) from 1997-2001, and then worked in the Computer Science Department at Georgia State University (Atlanta, USA) as an assistant professor from 2001 to 2004. At Georgia State University he collaborated with other faculty in joint research projects for developing a middleware for embedded software systems, web services, and bioinformatics. He later joined the Department of Computer Engineering at TOBB University of Economics and Technology (Ankara, Turkey) in 2004 and he is currently an associate professor and the head of the department since 2008. Dr. Dogdu’s research interests include database systems, software engineering, web services, and recently semantic web and web information systems in general. He is in the editorial boards of Turkish Journal of Electrical and Electronics Engineering and Computer Sciences, and International Journal of Knowledge-Based Organizations. He served as PI and Co-PI in several national and international research projects. He was awarded IBM Shared University Research Award in 2014 for a project in social media analytics. He is the director of IBM Big Data Analysis Laboratory at TOBB ETU. He has published more than 50 international conference and journal papers, and also two books in Turkish.