资源描述:
《Data challenges at Yahoo》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、DataChallengesatYahoo!RicardoBaeza-Yates&RaghuRamakrishnanYahoo!ResearchABSTRACTisrequiredtofurtherourunderstandingofhowcommunitiesformandevolve,howwecanfacilitatetheirinteractions,andhowweInthisshortpaperwedescribethedatathatYahoo!handles,thecanlearnfromsharedcommunityactivitiestoenhanceWebap-c
2、urrenttrendsinWebapplications,andthemanychallengesthatplications[12].WeviewYahoo!asanaturalfocalpointforthisthisposesforYahoo!Research.Thesechallengeshaveledtoevolution.thedevelopmentofnewdatasystemsandnoveldataminingtech-niques.TheWebasadeliverychannel.TheWebhasbecomeapow-erfulandubiquitousmean
3、sofdeliveringarangeofend-userand1.INTRODUCTIONcollaborativeapplicationstohundredsofmillionsofusers.Ason-lineapplicationdevelopmentmovesinthedirectionof“mash-ups”Yahoo!isauniverseofseveraldozensofdistinctpropertiesof-ofonlineAPIs(toawiderangeofcapabilitiesthatmaybecom-feringcapabilitiesrangingfro
4、memail,chat,onlinenews,shopping,binedintoanapplication),therequirementsonthebackendchangeandclassifieds;toWebsearchandsocialmediasites,suchasFlickrsubstantially.ThishascreatedaradicallydifferentapproachtoandYahoo!Answers.TheseWebsitesservemorethanhalfabil-developinganddistributingapplications,dis
5、ruptingthetraditionallionuniqueuserspermonthandgenerateseveralbilliondollarsofsoftwaredistributionmodel.Inturn,ithaschallengedustode-revenuethroughadvertisingandotherservices.Duetothisecosys-velopnewtypesofservice-orientedsoftwareplatforms,newkindstem,Yahoo!generatesoveradozenterabytesperdayofus
6、agedataofcustomizableapplicationenvironments,andforcedustothinkofvariousforms.JustthistypeofdataistheequivalentoftheentireaboutmassivelydistributedsystemswithnovelqualityofservicetextcontentoftheLibraryofCongress.guarantees,fail-overmechanisms,andtheabilitytomanagemas-Storingandmanagingthisocean
7、ofdataposesseveralimportantsivenumbersofapplicationinstances.challenges.Manyofthesechallengesdealwithdatamanagementandstretchcurrentlyavailabletechnology,motivatingthedevelop-TheWebaswisdom.TheWebhasbecomethelargestreposi-me