An Evolution-based Approach to Preserving User Preferences in Document-Category Management
Year: 2005
Type: article
Abstract: Document clustering is critical to automated document management, hereby a set of documents are clustered in multiple categories, each containing similar or relevant documents.Most previous research assumes time invariability of document category; i.e., not evolving over time after creation.The adequacy of an existing category understandably may diminish as it includes influxes of new documents over time, bringing about significant changes to its content.Following an evolution-based approach to preserving user preferences in document-category management, this study extends Category Evolution (CE) technique by addressing its inherent limitations.The proposed technique (namely, CE2) automatically re-organizes document categories while taking into account those previously established by the user.We empirically evaluate the effectiveness of CE2 in different document management scenarios that are created using a set of documents from Reuters.Our evaluation includes both CE and hierarchical agglomerative clustering (HAC) for performance benchmarks.Our analysis results show CE2 to be more effective than CE and HAC, showing higher clustering recall and precision.Our findings have interesting implications to research and practice, which are discussed together with our future research directions.
Source: AIS Electronic Library (AISeL)
Authors: Chih‐Ping Wei, Paul Jen‐Hwa Hu, Yen‐Hsien Lee
Institution: National Sun Yat-sen University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W1585165485
