| SIS Research Area - Data Management & Analytics
Research Theme
Central Concerns and Questions
We conduct research on exploratory search that involves users performing complex information seeking tasks to meet their research and investigative needs. Exploratory search tends to be iterative in nature, covering multiple facets of information as well as requiring much cognitive processing of intermediate search results. Such requirement is usually not well supported by the standard search engines.
Emerging Ideas and Initiatives
Our main idea is to develop a novel data model to analyze relationships among logical entities as the basic units of information in exploratory search. Our data model, known as TUBE, is designed to explore entities and their relationships derived from a collection of text or semi-structured documents using a series of search queries. At present, exploratory search tasks are not well supported by existing search engines due to the limitations of existing information retrieval models in modeling entity and relationship information. TUBE therefore provides multidimensional table structure to represent entity relationship networks in exploratory search and a set of operations to allow users to identify and refine interesting entities and relationships.
Selected Publications
[1] Zhen Sun, Ee-Peng Lim, Kuiyu Chang, Teng-Kwee Ong, Rohan Kumar Gunaratna. Event-Driven Document Selection for Terrorism. IEEE International Conference on Intelligence and Security Informatics (IEEE ISI-2005), Atlanta, Georgia, May 2005.
[2] Hady Wirawan Lauw, Ee-Peng Lim, Hweehwa Pang. TUBE (TextcUBE) for Discovering Documentary Evidence of Associations among Entities. ACM Symposium of Applied Computing (SAC2007), Seoul Korea, March 2007.
Collaborations and Industry Linkages
- Gary Marchionini, University of North Carolina at Chapel Hill
- Rohan Gunaratna, Nanyang Technological University
- Kuiyu Chang, Nanyang Technological University
- Hady Wirawan Lauw, A*Star/Now at Microsoft Research Search Technologies
|
