Enabling semantic search & retrieval with business-driven automated information extraction
-
PDF
432.5126953125Kb
- Author(s): M. Matton ; W. Bailer ; M. Elser ; J.P. Évain ; F. Glaeser ; A. Messina ; R. Mies ; F. Negro ; C. Ruiz Source: IBC2014 Conference, 2014 page ()
- Conference: IBC2014 Conference
- DOI: 10.1049/ib.2014.0041
- ISBN: 978-1-84919-927-8
- Location: Amsterdam, Netherlands
- Conference date: 11-15 Sept. 2014
- Format: PDF
Media production workflows become increasingly distributed and heterogeneous, requiring the use of multilingual content from networked media repositories. Many tasks of professionals in media production can be efficiently supported by automatic content analysis (e.g., speech recognition, quality analysis, action detection) and advanced semantic search and retrieval services. We present a system developed in the TOSCA-MP project, based on these technologies that enables a powerful semantic search & retrieval of audiovisual content. The system executes media analysis workflows using a flexible framework designed as a service oriented architecture. The TOSCA-MP architecture includes a task adaptive service orchestration based on business processes which handles automatic metadata extraction services with optional manual verification. The system also supports distributed heterogeneous repositories. We will briefly introduce these components. Finally, the system enables powerful new semantic search & visualization techniques, for which a dedicated semantic search engine has been created. We describe in detail the semantic search engine including its novel ways of content visualization and navigation. We also show how the TOSCA-MP architecture efficiently supports these user interfaces. Further, we discuss the feedback received by professional users who have successfully tested the system during evaluation workshops.
Inspec keywords: meta data; business data processing; service-oriented architecture; data visualisation; query processing
Subjects: Software engineering techniques; Information networks; Information retrieval techniques; Business and administrative computing; Graphics techniques
Related content
content/conferences/10.1049/ib.2014.0041
pub_keyword,iet_inspecKeyword,pub_concept
6
6