All Title Author
Keywords Abstract

Publish in OALib Journal
ISSN: 2333-9721
APC: Only $99


Relative Articles


Faceted Subtopic Retrieval: Exploiting the Topic Hierarchy via a Multi-modal Framework

DOI: 10.4304/jmm.7.1.9-20

Keywords: subtopic retrieval , multi-modal analysis , search result clustering , hierarchical topic model , social media , connected multimedia

Full-Text   Cite this paper   Add to My Lib


The overwhelming amount of web videos posted on the social mediawebsites make effective browsing and search a challenging task. Theuser-provided metadata, has been proved useful in large-scale videoorganization and retrieval. Search result clustering, which utilizesthe associated metadata to cluster the returned results intosemantic groups according to its involved subtopics, has shown itsadvantages. Most of the existing works on search result clusteringare devoted to solving the ambiguous problem resulted fromgeneral queries. In this paper, we propose the problem of extit{faceted subtopic retrieval}, which focus on more complexqueries concerning political and social events or issues.Hierarchical topic model (hLDA) is adapted to exploit the intrinsictopic hierarchy inside the retrieved collections. Furthermore, thispaper offers a new perspective of multi-modal video analysis byexploring the pairwise visual cues deriving from duplicate detectionfor constrained topic modeling. We modify the standard hierarchicaltopic model by integrating: 1) query related Supervision knowledge(ShLDA) and 2) duplicate Relation constraints (RShLDA). Carefullydesigned experiments on web-scale video dataset validate theproposed method.


comments powered by Disqus

Contact Us


WhatsApp +8615387084133

WeChat 1538708413