Similarity-Based Cross-Media Retrieval for Events

Piroska Lendvai; Thierry Declerck

In: Ralph Bergmann; Sebastian Görg; Gilbert Müller (Hrsg.). Proceedings of the LWA 2015 Workshops: KDML, FGWM, IR, and FGDB. GI-Workshop-Tage "Lernen, Wissen, Adaption" (LWA-2015), October 7-9, Trier, Germany, CEURS, 10/2015.


Our goal is to link social media content to contextually relevant information in complementary media in the domain of daily news. Web links from tweets with user-included URLs are transferred to URLless tweets, using manually annotated events. The new cross-media ties establish authoritative feedback documents for unsupported social media content, and enable extracting an improved set of event-denoting terms based on longest common subsequences between tweets and documents.


ir2015_crc_64_lendvai_declerck.pdf (pdf, 171 KB )

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence