多媒体网络的信息提取：开源平台及试验台（英文）.pdf

Int J Multimed Info Retr (2014) 3:97–111 DOI 10.1007/s13735-014-0051-2 REGULAR PAPER Information extraction from multimedia web documents: an open-source platform and testbed David Paul Dupplaw · Michael Matthews · Richard Johansson · Giulia Boato · Andrea Costanzo · Marco Fontani · Enrico Minack · Elena Demidova · Roi Blanco · Thomas Griffiths · Paul Lewis · Jonathon Hare · Alessandro Moschitti Abstract The LivingKnowledge project aimed to enhance Keywords Multimedia retrieval · Web analysis · the current state of the art in search, retrieval and knowl- Text analysis · Opinion analysis · Image analysis · edge management on the web by advancing the use of sen- Open-source software timent and opinion analysis within multimedia applications. To achieve this aim, a diverse set of novel and complementary 1 Introduction analysis techniques have been integrated into a single, but extensible software platform on which such applications can be built. The platform combines state-of-the-art techniques Most documents on the web are now inherently multimedia. for extracting facts, opinions and sentiment from multime- Improving the retrieval of multimedia documents in terms dia documents, and unlike earlier platforms, it exploits both of precision in response to more demanding queries requires visual and textual techniques to support multimedia informa- improvements in knowledge extraction and indexing from tion retrieval. Foreseeing the usefulness of this software in multimedia data. This is a particularly challenging task as the wider community, the