Ecole d'ingénieur et centre de recherche en Sciences du numérique

Event-based cross media question answering

Liu, Xueliang; Huet, Benoit

Multimedia Tools and Applications, June 2014, ISSN: ISSN 1380-7501

User generated content, available in massive amounts on the Internet, is receiving increased attention due to its many potential applications. One of such applications is the representation of events using multimedia data. In this paper, an event-based cross media question answering system, which retrieves and summarizes events on a given topic is proposed. In orther words, we present a framework for leveraging social media data to extract and illustrate social events automatically on any given query. The system is built in three steps. First, the input query is parsed semantically to identify the topic, location, and time information related to the News of interest. Then, we use the parsed information to mine the latest and hottest related News from social news web services. Third, to identify a unique event, we model the News content by latent Dirichlet Allocation and cluster the News using the DBSCAN algorithm. In the end, for each event, we retrieve both textual and visual content of News that refer the same event. The resulting documents are shown within a vivid interface featuring both event description, tag cloud and photo collage.

Document Doi Bibtex

Titre:Event-based cross media question answering
Mots Clés:events, social media, illustration, cross media, question answering
Département:Data Science
Eurecom ref:4347
Copyright: © Springer. Personal use of this material is permitted. The definitive version of this paper was published in Multimedia Tools and Applications, June 2014, ISSN: ISSN 1380-7501 and is available at :
Bibtex: @article{EURECOM+4347, doi = {}, year = {2014}, month = {06}, title = {{E}vent-based cross media question answering}, author = {{L}iu, {X}ueliang and {H}uet, {B}enoit}, journal = {{M}ultimedia {T}ools and {A}pplications, {J}une 2014, {ISSN}: {ISSN} 1380-7501}, url = {} }
Voir aussi: