PhD Dissertation Defense by CHONG Wen Haw | Context Recovery in Location-based Social Networks

Context Recovery in Location-based Social Networks

About The Talk

Location-Based Social Networks (LBSN) are platforms where users post content from various locations. With this general LBSN definition, many existing social media platforms that support user-generated content using mobile devices qualify as LBSNs. For these platforms, we seek to recover the venue and semantic contexts of user posts. Such information is useful for user profiling and to support various applications such as venue recommendation and location-based advertising.

For venue context recovery, we focus on Twitter where the venue information is often missing. We frame the problem as fine-grained geolocation whereby we geolocate tweets to their specific posting venues such as a restaurant, a shop etc. In this talk, we present our geolocation work focusing on tweets from users without location history; and on tweets contained in sequences posted by the same user within a short time interval. For semantic context recovery, we focus on the entity linking problem, which seeks to recover the entity mentioned or implied in a short content post. We present our work on implicit entity linking which aims to identify local cuisines in food-related posts. We link such posts directly to food entities in a knowledge base without the need for mention extraction.

Speaker Biography

CHONG Wen Haw is a PhD candidate in the School of Information Systems, Singapore Management University, working with Professor Ee-Peng Lim. He received his Bachelor of Engineering (Electrical), and Master of Science (Statistics) from the National University of Singapore in 2003 and 2007 respectively. His current research focuses on tweet geolocation and entity linking in location based social media.