Title of article :
Browsing Image Collections with Representations of Common-Sense Activities
Author/Authors :
Gordon، Andrew S. نويسنده ,
Issue Information :
ماهنامه با شماره پیاپی سال 2001
Abstract :
To support browsing-based subject access to image collections, it is necessary to provide users with networks of subject terms that are organized in an intuitive, richly interconnected manner. A principled approach to this task is to organize the subject terms by their relationship to activity contexts that are commonly understood among users. This article describes a methodology for creating networks of subject terms by manually representing a large number of common-sense activities that are broadly related to image subject terms. The application of this methodology to the Library of Congress Thesaurus for Graphic Materials produced 768 representations that supported users of a prototype browsing-based retrieval system in searching large, indexed photograph collections.
Keywords :
relational learning , Naive Bayes , text categorization , predicate invention
Journal title :
Journal of the American Society for Information Science and Technology
Journal title :
Journal of the American Society for Information Science and Technology