Publication Framework for Content Based Search in Heterogeneous Distributed Multimedia

Michel Crucianu

Description

Niveau: Supérieur, Doctorat, Bac+8
Publication Framework for Content-Based Search in Heterogeneous Distributed Multimedia Databases Franc¸ois Boisson, Michel Crucianu, Dan Vodislav Vertigo - CEDRIC & Wisdom, CNAM 292 rue St. Martin, 75141 Paris cedex 03, France 10 September 2008 Abstract Search for multimedia content by the large public is expected to ex- perience at least two major evolutions in the near future: addition of content-based criteria to those issued from textual metadata and signifi- cant increase in the volume of digitized content available at each provider. In this context, most providers will need a local index in order to search in their own database. We argue that the local indexes are unlikely to be all of the same type and that many providers want full control over their local index. To support these requirements in a flexible and scalable way, we outline a framework based on the publication by every provider, in a consensual format, of representatives associated to those parts of the local index where the provider intends to make content “findable” by oth- ers. The representatives published by all the providers serve to build a global distributed index employed for answering content-based similarity queries. 1 Introduction The evolution of user requirements and of enabling technologies will have a sig- nificant impact on how online search for multimedia content is performed. A major demand of the users is to avoid the “keyword bottleneck”, in part by the inclusion of content-based search criteria [21].

• cbmr paradigm


• global distributed


• timedia documents


• large institutional


• them can


• multimedia content


• based multimedia retrieval


• keyword bottleneck

Informations

• Publié par: profil-zyak-2012
• Nombre de lectures: 29
• Langue: English

Extrait

PublicationFrameworkforContent-BasedSearch inHeterogeneousDistributedMultimedia Databases Franc¸oisBoisson,MichelCrucianu,DanVodislav Vertigo-CEDRIC&Wisdom,CNAM 292rueSt.Martin,75141Pariscedex03,France 10September2008 Abstract Searchformultimediacontentbythelargepublicisexpectedtoex- perienceatleasttwomajorevolutionsinthenearfuture:additionof content-basedcriteriatothoseissuedfromtextualmetadataandsignifi- cantincreaseinthevolumeofdigitizedcontentavailableateachprovider. Inthiscontext,mostproviderswillneedalocalindexinordertosearch intheirowndatabase.Wearguethatthelocalindexesareunlikelyto beallofthesametypeandthatmanyproviderswantfullcontrolover theirlocalindex.Tosupporttheserequirementsinaflexibleandscalable way,weoutlineaframeworkbasedonthepublicationbyeveryprovider, inaconsensualformat,of representatives associatedtothosepartsofthe localindexwheretheproviderintendstomakecontent“findable”byoth- ers.Therepresentativespublishedbyalltheprovidersservetobuilda globaldistributedindexemployedforansweringcontent-basedsimilarity queries. 1Introduction Theevolutionofuserrequirementsandofenablingtechnologieswillhaveasig- nificantimpactonhowonlinesearchformultimediacontentisperformed.A majordemandoftheusersistoavoidthe“keywordbottleneck”,inpartby theinclusionofcontent-basedsearchcriteria[21].Butindexingthecontentof images,videosormusicisverydifferentfromindexinghypertext,bothbecause ofthenatureandvolumeofthedataandbecauseofspecificrightsissues.To clarifytheimplicationsoflargescalecontent-basedsearchindistributedrepos- itoriesandhighlightsomeresearchchallengesthatwedeemimportantinthis context,westartbytakingacloserlookattwomajortypesofexistingorpo- tentialmultimediacontentproviders:thegeneralpublicandlargeinstitutional archives. 1