TUBE (Text-cUBE) for discovering documentary evidence of associations among entities

• Main Authors: LAUW, Hady, LIM, Ee Peng, PANG, Hwee Hwa
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2007
• Subjects: association discovery; interactive IR; Databases and Information Systems; Numerical Analysis and Scientific Computing
• Online Access: https://ink.library.smu.edu.sg/sis_research/372; https://ink.library.smu.edu.sg/context/sis_research/article/1371/viewcontent/Tube__Textcube__for_Discovering_Documentary_Evidence_of_Association_among_Entities__edited_.pdf
• Institution: Singapore Management University

User-driven discovery of associations among entities, and documents that provide evidence for these associations, is an important search task conducted by researchers and do-main information specialists. Entities here refer to real or abstract objects such as people, organizations, ideologies, etc. Associations are the inter-relationships among entities. Most current works in query-driven document retrieval and finding representative subgraphs are ill-suited for the task as they lack an awareness of entity types as well as an intuitive representation of associations. We propose the TUBE model, a text cube approach for discovering associations and documentary evidence of these associations. The model consists of a multi-dimensional view of document data, a flexible representation of multi-document summaries, and a set of operations for data manipulation. We conduct a case study on real-life data to illustrate its applicability to the above task and compare it with the non-TUBE approach.