meeting of the minds 2012
DESCRIPTION
Meeting of the MindsTRANSCRIPT
C A R N E G I E M E L L O N U N I V E R S I T Y I N Q A T A R
Meetingofthe Minds
Meeting of the minds is an annual symposium at Carnegie Mellon
University that gives students an opportunity to present their research
and project work to a wide audience of faculty, fellow students, family
members, industry representatives and the larger community. Students
use posters, videos and other visual aides to present their work in a
manner that can be easily understood by both experts and non experts.
Through this experience, students learn how to brindege the gap between
conducting research and presenting it to a wider audience. A review
committee consisting of industry experts and faculty members from other
universities will review the presentations and choose the best projects and
posters. Awards and certificates are presented to the winners.
Table of Contents
POSTER # TITLE PAGE
Business Administration Poster Q1 Islamic Finance Meets Wall Street 1
Computer Science Posters Q2 Building a Virtual Computer from the Ground Up 3
Q3 DevelopingScenariosforaQatar-specificRoadSafetySimulator 5
Q4 EvaluationoftheAbilityofaRobottoEmbodyDifferentCulturalTraits 7
Q5 EvaluationofVariationsinGivingDirectionsAcrossCultures 9
Q6 Image Processing on the Cloud: Characterizing Edge Detection on Biomedical Images 11
Q7 MalwareInc.–FacebookandGoogleAppEngine 13
Q8 MalwareInc-WebBrowsers 15
Q9 Multi-RobotSimulation 17
Q10 ProjectingNamedEntityBoundariesfromEnglishtoArabic 19
Q11 SCOUT:ExtendingtheReachofSocial-BasedContext-AwareUbiquitousSystem 21
Information Systems Posters Q12 A3 (A-Cubed) 23
Q13 EZIntern:InternshipTrackingSystem 25
Q14 Lost&Found 27
Q15 MoltaQatartans:TartansForumSystem 29
Q16 UsingMobileTechnologyforEnhancingYoungQatariHealthBehavior 31
Humanities and Social Sciences Q17 Gettin’theFlow;Makin’GoodGrades 33
Q18 ServiceLearningatCMU-Q:Motivations,Gains,andChallenges 35
Post-Graduate Posters QG1 ChallengesinMobileOpportunisticNetworks 37
QG2 CharacterizationofHadoopMapReduceApplications 39
QG3 CoGRS:ACenter-of-GravityReduceTaskScheduleforMapReduce 41
QG4 GreenLoc:EnergyEfficientWi-Fi-basedIndoorLocalization 43
QG5 Hala2.0:ConsiderationsforDevelopingaTestBedforMulti-Lingual,
Cross-CulturalHumanRobotInteraction 45
QG6 PerformancePredictionofMapReduceApplicationsinElasticComputeClouds 47
QG7 SmartReader:Anaturallanguageprocessing-basedactiveandinteractivesystem
foraccessingEnglishlanguagecontentandadvancedlanguagelearning 49
QG8 VOtus:AFlexibleandScalableMonitoringFrameworkforVirtualizedClusters 51
Islamic Finance Meets Wall Street
AuthorEdmond Abi Saleh (BA 2011)
Faculty AdvisorPatrick Sileo, Ph.D
CategoryBusiness Administration
AbstractThis research explores the feasibility of establishing Islamic banking and finance within the capitalist
Americanfinancialsystem.This isa relevantandpressing topicseeingasMuslimcommunitieshave, for
centuries, slowlyandsurelyestablished themselveswithinAmericansociety. It is thus interesting to look
at theirfinancialparticipation in theAmericanmarketespecially in thewakeof theeconomiccrisis.More
precisely,howtheAmericaneconomyversustraditional Islamicfinancehasfacedthecrisisandhowthis
potentiallyaffectscapitalismintheUnitedStates.Inthatregard,theresearchwillarguethat:Islamicfinance
(throughcanonicaltextsandtheirapplicationwithinMuslimcommunities)isonlypartlyinaccordancewith
theAmericancapitalisteconomicpractices,whichmeansthatitcannotbeimplementedintheUnitedStates
onalargescaleorinapreciseandcorrectmanner(meticulouslyfollowingIslamiclaw).Theresearchwillput
forththisargumentbyfirstexamininghowMuslimcommunitiescametoAmericaandhowwell-established
orinfluentialtheyare.Second,itwillanalyzewhatthesespecificcommunities’religiousvaluessayaboutthe
economicsphereandhowitshouldoperate.Finally,allthisinformationwillbebroughttogethertolookat
presentdayIslamicfinanceintheUnitedStates.Theobjectiveistoassesshowcompatibleitcanbewith
theAmericancapitalistvaluesarguingthatthosetwofinancialmodelsaretoodivergentintheiressencesto
extensivelyco-existatpresent.Inasecondpart,theresearchwillanalyzetheIMXL,whichistheDowJones
Islamicmarketindex.TheanalysiswilltrytolookforsimilaritiesorgapsbetweenIslamicmarketsandregular
markets. Thisquantitative analysiswill reveal that there is little or nodifferencebetween theDowJones
IslamicmarketindexandtheregularDowJonesIndex.Theanalysiswillalsosuggesthypothesesastothe
behavioroftheIslamicindexinthemarket.
1
Building a Virtual Computer from the Ground Up
AuthorsKenrick Fernandes (CS 2014)Jyda Moussa (CS 2014)
Faculty AdvisorKemal Oflazer, Ph.D.
CategoryComputer Science
Abstract:Inthiswork,wepresentourexplorationofunderstandingandbuildingafull-fledgedcomputersystemfrom
scratch–fromthegatelevelallthewaytothesoftwarelevels.Weimplementedthiscomputerusingthetools
containedintheElementsofComputingSystemsframework.Forthisindependentstudy,weweremotivated
toobtainabetterunderstandingofhowamoderncomputerfunctions,fromthehardwarelevelrightupto
theinterfacesthatusersworkwith.Buildingavirtualcomputerfromthegatelevelrequiredustobuildthe
individualgatecomponents,followedbyanArithmeticLogicUnit(ALU)andfinallyaCentralProcessingUnit
(CPU).Wethenimplementedanassemblertotranslateassemblylanguageprogramstomachinelanguage
thatwouldrunonthecomputerwebuilt.Ourposteraimstodescribeourexperienceinthis independent
studyaswellastheindividualprojectswehaveimplementedtodate.
3
Developing Scenarios for a Qatar-specific Road Safety Simulator
AuthorRaggi al Hammouri (CS 2012)
Faculty AdvisorBrett Browning, Ph.D.
CategoryComputer Science
Abstract:Driving simulators are increasingly being utilized as training devices to complement traditional in-vehicledriver training.Byallowingscenario-basedcoachingand tailoring the instructivecontent to thestudent’slearningneeds,drivingsimulatorscanplayasignificantroletowardsimprovingroadsafety.
WilliamsTechnologyCentreinQatarhasstartedaprojecttodesignandbuildacustomizeddrivingsimulationplatformfortheadvancedtrainingofdriversinQatar.Thisworkfocusesonthecreationoftheprogramsthatspecifythescenario,andcontrolthebehaviorofthesimulatedvehiclesintheWilliams’drivingsimulator.WeaimtodeveloparichenvironmentwheresubjectdriversencounterrealisticmodelingofdrivingbehaviorasfoundinQatar,andcontrolledpresentationofeventsandsituations.
WehavecreatedaroadandanumberoftrafficmodelsforDoha,aswellasspecifictrainingscenarios.WedescribethearchitectureofthescenariosubsystemofWilliams’drivingsimulator,focusingontherepresen-tationofthevirtualworldandthecontrolofvehiclebehavior.Wepresentsomeexamplescenariosandshowtheresultsofthesescenariosinpractice.
5
Evaluation of the Ability of a Robot to Embody Different Cultural Traits
AuthorAmna AlZeyara (CS 2014)
Faculty AdvisorsMajd F. Sakr, Ph.D.Micheline Ziadee
Abstract:Ourfocusisthedevelopmentofeffectivemulti-lingual,cross-culturalHuman-Robot-Interaction.Inthiswork,
weattempttounderstandthedifferentvisualaccentsinArabicandAmericanfacialexpressionsandcreate
culture-specificfacialexpressionsforafemalemulti-lingual,cross-culturalrobottestbed.
Ourworkistwofold:theidentificationoftheexistenceofaccentvariationinfacialexpressionsacrosscul-
turesandthevalidationofhumanrecognitionofthesesaccents.Facialexpressionsareembodiedinculture
andarecrucialforeffectivecommunication;hencetheyplayanimportantroleinmulti-lingualcross-cultural
Human-Robot-Interaction.ElfenbeinandAmbadyfoundthattherearedifferentaccentsinfacialexpressions
whichareculture-specificand that thedifferences inexpressionsbetweenculturescancreatemisunder-
standings.SeveralstudiescomparedAmericanexpressionswithexpressionsfromotherculturesbutnoneof
themincludedArabicfacialexpressions.
Inthiswork,weutilizetwosetsoffacialexpressions;onespecifictotheArabiccultureandtheotherspe-
cifictotheAmericanculture.WebasetheAmericanexpressionsonsamplesfromMMIFacialExpression
Database,aweb-baseddatabaseforfacialexpressionanalysis.However,nosuchdatabaseexistsforArabic
facialexpressions.Consequently,werecordedvideosofyoungArabwomennarratingstoriesthatexpress
sixemotions:happiness,sadness,surprise, fear,disgust,anddisappointment.Thesevideosareanalyzed
to extract Arabic accents in facial expressions. Accents frombothArab andAmerican cultures are then
implementedona3Dmodel.Wevalidatetheexistenceoftheseaccentsbyinvestigatingwhetherhumans
candetectdifferencesbetweenArabicandAmericanfacialexpressions.Ourinitialobservationshaveshown
differencesinangerexpressions.Arabicexpressionofangerischaracterizedbyraisedeyebrowswhilethe
Americanexpressionofanger,asdescribedbyPaulEkman,ishighlightedbylowered,drawn-togethereye-
brows.Ournextstepwillinvolvecreatingculture-specificlinguisticcontentandpairingitwithculture-specific
expressions.
7
Evaluation of Variations in Giving Directions Across Cultures
AuthorsHuda Gedawy (CS 2012)Micheline Ziadee (Research Assistant)
AdvisorMajd F. Sakr, Ph.D.
CategoryComputer Science
AbstractThisworkexploresthedifferencesindirection-givingstrategiesbetweentwogroups,nativeArabicandnativeEnglishspeakers.Thisstudywillhelpinfluencedesigndecisionsformulti-lingual,cross-culturalhumanrobotinteractions.
Thereareclearculturalinfluencesonmodesofcommunication.Previousresearchstudiesfoundthatdirection-givingtechniquesandstrategiesvarybetweendifferentculturalgroups.BurhanudeencomparedJapaneseandEnglishnativespeakersandfoundthatlocatorremarksaremorefrequentlyusedbyJapanesenatives,whiletheuseofdirectivesismorecommonwithnativeEnglishspeakers.
Inthiswork,weexaminethediscoursefornavigationinstructionsofmembersoftwotargetgroups,ArabicnativespeakersandEnglishnativespeakers.Weaddressthefollowingquestions:Howdoeslanguageandstrategiesusedforprovidingdirectionsvarybetweenthesetwogroups?Whatarethedifferencesandwhatarethesimilarities?Arethereanypossiblegender-relateddifferencesingivingdirections?
Werecorded56participantsgivingoraldirectioninstructionsfor3specificlocationsintheCarnegieMellonQatarcampus.Wetranscribedtheoralrecordingsandannotatedthetextsbycategorizingthedirectionalexpressions into: landmarks, units (time and distance), cardinals, movement commands, conditionalmovement commands, advisory commands and positional commands. Non-directional commands, likeintermediatedetails,werecategorizedasfillers.Ouranalysisalsoincludedthenumberofpauses,repetitionanderrorcorrection.
Ourresultsshowedthatwhengivingdirections,EnglishnativespeakersuseaconsiderablyhighernumberofwordsthanArabicnativespeakers.Also,thedifferenceinnumberofwordsbetweenmalesandfemalesinthefirstgroupismuchhigherthanthatinthesecondgroup.Englishnativespeakershadahigherfrequencyofusingcardinals,intermediateinformation,andfilledpauses. Ontheotherhand,ArabicnativespeakersdifferedgreatlyfromEnglishnativespeakersintheiruseofunitsofdistanceandrepeatedinformation.Basedontheseresults,weconcludethatculture,language,andgender
influenceaspeakersdiscourseandstrategyforgivingdirections.
9
Image Processing on the Cloud: Characterizing Edge Detection on Biomedical Images
AuthorManoj Reddy (CS 2013)
Faculty AdvisorsMohammad HammoudMajd F. Sakr, Ph.D.
CategoryComputer Science
AbstractInordertoanalyzeanddeducevaluableinformationfrombigimagedata,wehavedevelopedaframeworkfordistributedlarge-scaleimageprocessinginHadoopMapReduce.Avastamountofscientificdataisnowrepresentedintheformofimagesfromsourcesincludingcameras,medicaltomographyandastronomicalremotesensing.Applyingalgorithmsontheseimageshasbeencontinuallylimitedbytheprocessingcapacityofasinglemachine.MapReducecreatedbyGooglepresentsapotential solution.MapReduceefficientlyparallelizescomputationbydistributingtasksanddataacrossmultiplemachines.Hadoop,anopensourceimplementationofMapReduce,isgainingawidespreadpopularityduetofeaturessuchasscalability,faulttoleranceandability tousecommodityclusters.Hadoop isprimarilyusedwith text-based inputdata. Incontrast,itsabilitytoprocessimageshasnotbeenfullyexplored.WeproposeaframeworkthatefficientlyenablesimageprocessingonHadoop.
Existingapproachesindistributedimageprocessingsufferfromtwomainproblems:(1)inputimagesneedtobeconvertedtoacustomfileformatand(2)imageprocessingalgorithmsrequireadherencetoaspecificAPI thatmight imposesome restrictionsonapplyingsomealgorithms toHadoop.Our frameworkavoidstheseproblemsby:(1)bundlingallsmallimagesintoonelargefilethatcanbeseamlesslyparsedbyHadoopand(2)relaxinganyrestrictionbyallowingadirectportingofanyimageprocessingalgorithmtoHadoop.AReduce-lessjobisthenlaunchedwherethecodeforprocessingimagesandamechanismtowritethembackseparatedtoHDFSare includedinMappers.Wehavetestedtheframeworkwithastate-of-the-art imageprocessingalgorithm,EdgeDetection,ona largedataset,3760,ofbiomedical images.WeobservedthatusingHadoopSequenceFilestobundlealargenumberofsmallimagesintoalargesequencefilemightnotbethemostefficientsolution,asweobservedastorageoverheadofupto3x.Furthermore,andtoexamineHadoop’sbehaviorwithimageprocessing,wearecharacterizingEdgeDetectionalongseveraldimensions,suchasdegreeofparallelismandnetworktrafficpatterns.Ourcharacterizationstudyhasshownthatvaryingthenumberofmap taskshasasignificant impactonHadoop’sperformance.Thebestperformancewasobtainedwhenthenumberofmaptasksequals thenumberofavailableslots,as longas theapplicationresource demand is satisfied. Additionally, we observed a speedup of 2.1X as compared to the defaultHadoopconfiguration.
11
Malware Inc. – Facebook and Google AppEngine
AuthorsTalal Al-Haddad (CS 2013)Manoj Reddy (CS 2013)
Faculty AdvisorThierry Sans, Ph.D.
CategoryComputer Science
AbstractAnewgenerationofsoftwarehasemergedwiththeincreasingpopularityofwebandcloud-basedapplications.
CloudcomputingplatformssuchasFacebookandGoogleAppEngineallowthirdpartydeveloperstobuild
applications forusers that runona remote infrastructure.Thesenewplatformsgive rise tonewsecurity
threatsandanewgenerationofmalware.Malware,short formalicioussoftware, issoftwaredesigned to
disruptcomputeroperation,gathersensitiveinformation,orgainunauthorizedaccesstocomputersystems.
Inthiswork,weaimtoevaluatethesecurityriskofexposuretomalwareonthetwopopularcloudapplication
ecosystems:FacebookandGoogleAppEngine.
Google AppEngine applications are hosted on Google infrastructure and can access Google services and also
user’spersonaldatasuchGmailandGoogleCalendar.Asaproofofconcept,wehavedevelopedamalware
thatpretendstodousefulstatisticsontheuser’smailboxbutinfact,thisapplicationsearchesthroughemails
forlogins,passwordsandcreditcardinformation.Whenfound,thesedataaresenttoaremotewebserver
controlledbytheattacker.Thisoperationcannotbedetectedbytheuser.
Facebookallowsdeveloperstowriteapplicationsthatcanperformawidevarietyoftaskssuchascollaborative
gaming and access user information for social statistics.Contrary toGoogle, these applications are not
storedonFacebookserversbutonserverscontrolledbythedeveloper.Asaconsequence,whenusersuse
suchapplications, theirprofiledatacanbestolenandsent toaremotewebserversimilar to thecaseof
GoogleAppEngine.
Themalwareapplicationsdemonstratethat it iseasytodeploymaliciousapplicationsoncloudplatforms
andthisisahugecauseforconcernsinceeverincreasingamountofsensitivedataisstoredonline.Future
workincludesdevelopingtechniquesforautomatedprofilingofapplicationstosensemaliciousactivityina
sandboxedenvironment.
13
Malware Inc - Web Browsers
AuthorsBaljit Singh (CS 2014) Fahim Dalvi (CS 2014)
Faculty AdvisorThierry Sans, Ph.D.
CategoryComputer Science
AbstractAnextensionisanapplicationthatenhancesthefunctionalityofabrowser.Thedifferenttypesofextensions
available today range from includingacalculator inabrowser tostoringpasswords in thebrowser inan
encrypted form.Thegoal of theMalware Incprojectwas toevaluate the riskof exposure tomalware in
popularbrowserssuchasGoogleChromeandMozillaFirefox.AMalwareisanysoftwarethatmaliciously
triestodisruptcomputeroperations,stealsensitiveinformationandsoon.Wewereabletocreateseveral
malwareintheformofextensions,twoofwhicharedescribedbelow.
InGoogleChrome,wecreatedanextensionthatcouldrecordthepasswordsofuserswhentheytrytologin
usinge-bankingmode.Whentheuserentersdatausinganon-screenkeyboard,ascreenshotofthekeyboard
wascapturedwithemphasisonthebuttonthatwaspressed.Usingtheseriesofscreenshotsreceived,the
developercouldthendeducethedataenteredeasily.
OnMozillaFirefox,wewereabletocreateanextensionthatcouldremotelydownloadandexecuteexternal
programs(whichloggedallthedatathatyoutypedonyourkeyboard),andthenreadfiles(thelogs)fromthe
user’ssystemandtransmitthembacktothedeveloperwithoutanyextrapermissions.
Theantivirusisunabletotrackthiscommunication,asitisentirelythroughthebrowser,oneofthetrusted
softwareonthemachine.Fromtheinformationwehavegatheredthroughthisproject,ouraimistocreate
newtechnologiesthatcandetectandstopmaliciousextensions.Oneofourideasistohaveanextension
thatcankeeptrackofallthecommunicationinitiatedbyotherextensions,essentiallycreatingan“antivirus”
forthebrowser.
15
Multi-Robot Simulation
AuthorsSidra Alam (CS 2013)Hanan Alshikhabobakr (CS 2013) Shailja Relwani (CS 2014)
Faculty AdvisorsM. Bernardine Dias, Ph.D. Brett Browning, Ph.D.
CategoryComputer Science
AbstractMulti-robot research can be very challenging due to the difficulty of operating several complex vehicles
simultaneously. Simulation can be a powerful tool to speed up software development if the simulator is
sufficientlyaccurate.Theprojectbuildsfromanexistingrobotsimulationtool–theUrbanSearchandRescue
Simulator(USARSim).ThefocusoftheresearchwastocustomizeandenhancetheUSARSIMmulti-robot
environment to enablehighconfidenceof capturing thecomplexitiesof real robot execution, specifically
for the robots in theQri8 robotics lab atCarnegieMellonUniversity inQatar. In terms of an application
domain, our focuswas on disaster response scenarios. The primary task for the robots in this scenario
wasexploration.For the simulation testing,wecreatedmaps for thedisaster response scenario andwe
addedphysicalconfigurationmodelsfortherobots.WeusedgenericrobotcontrolsoftwarecalledPlayerto
controltherobotsinthesimulator.Tovalidatetheexperiment,weranatestwithrealrobotsandanidentical
simulationtestandcomparedtheexecutedbehaviors.
17
Projecting Named Entity Boundaries from English to Arabic
AuthorNehal Elkady Hussein (CS 2013)
Faculty AdvisorBehrang Mohit, Ph.D.
CategoryComputer Science
Abstract: NamedEntityRecognition(NER)istheprocessofidentifyingspansoftextthatconstitutepropernamesandthenclassifyingthemaccordingtotheirtype(e.g.namesofpersons,locations,organizations,etc.).Englishisoneofthefewlanguageswhichhavedecentamountsofhuman-annotateddatafortrainingNERsystems.Incontrast,languageslikeArabicdonothaveenoughannotateddataandrobustNERsystems.ToovercomethisresourceshortageofArabic,weimplementedsoftwarethatautomaticallytagsArabictextwithnamedentityinformationusingacross-lingualprojection.Namedentityprojectionisaframeworktousewordalign-mentstotransferthenamedentityknowledgefromaresourcerichlanguage(e.g.English)toaresource-poorlanguage(e.g.Arabic).
Aword-alignedparallelcorpuscontainssentencesinEnglish,theirtranslationinArabic,andtheword-levelalignmentinformationbetweentheparallelsentences.OurnamedentityprojectionsoftwareusesthewordalignmentstofindtheproximityofthenamedentitiesontheArabicsentence.OutputoftheprojectionsystemistheannotationofthenamedentitiesontheArabicsideoftheparallelcorpus.Majorchallengesinourpro-jectioncomefromthedivergenceofthelanguagesandthenoiseinthewordalignments.Also,sinceArabicwordscouldbealignedtomorethanoneEnglishword,andviceversa,weobservecaseswhereoneArabicwordgetsmappedtotwoEnglishentities.
Weintroducedseveralheuristicsinourprojectionalgorithm.Twoimportantonesare:(1)Welabeledanamedentityifandonlyifithasagapoftwowordsorless,andthewordsformingthegaparenotalignedtootherwordswithinthesamesentence; (2) Ifanyoftheelementsconstitutinganamedentitywererepeatedweremovethemandthenlabelthenamedentity.
Theevaluationofourprojectionsystemwasbasedoncomparingitsoutputwiththegoldstandardanswersprovidedbyhumanannotation.Todothat,weselectedthirtyrandomArabicsentencesfromtheparallelcor-pusandannotatedthemwiththegoldstandardnamedentityboundaries.Wethencomparedtheresultsofourprojectionoutputagainstthisgold-standarddata.Thedisagreement(error)betweentheprojectionoutputandthegold-standarddatawasabout6.3%.Moreover,weusedthestandardprecisionandrecallmetricstoevaluatetheprojection.Precisionmeasurestheaccuracyoftheprocedureandrecallmeasuresthecoverageofit.Forourthirtysentencescorpus,weachievedaprecisionof81.4%andarecallof76%whicharepromis-ingbaselinesforcreatingnewnamedentityresourcesforArabic.Furthermore,weusedourprojecteddatatoenhanceabaselinenamedentitysystemandachievedasignificant4%averageimprovement.
19
SCOUT: Extending the Reach of Social-Based Context-Aware Ubiquitous System
AuthorDania Abed Rabbou (CS 2012)
Faculty AdvisorsAbderrahmen Mtibaa, Ph.D. Khaled Harras, Ph.D.
CategoryComputer Science
AbstractTheproliferationofsocial-networks,localizationsystems,andhigh-endmobiledeviceshascreatedafertile
groundforthedevelopmentofsystemsthatareawareofinterestsandadaptivetolocation.Withtheburgeon-
ingofthedomainofsocial-basedcontext-awaresystems,numerouschallengesarebecomingofincreased
importance.Onesuchchallenge,notaddressedsofarbytheresearchcommunity,isend-to-endcommuni-
cationbetweenubiquitoussystemsandtheirusers.Communicationinexistingsystemsiseithercentralized
ordistributed.Centralizedsystemsrequireuserstobeconnectedtoaserveratalltimesandthusassume
theavailabilityofinternetconnectivityeverywhere.Inreality,internetconnectivitymaybeabsent,charged,
energy-consuming,heterogeneous,andover-loaded.Asanalternative,distributedcommunicationenables
userstoobtaininformationfromneighboringdevicesbuttheavailabilityofinformationandtheextentofits
disseminationaredictatedsolelybyusermobilityandcontacts.
Werealize theneedforanewhybridmodethat leveragesthecentralizedanddistributedcommunication
modes.Usingthehybridmode,aubiquitoussystemdisseminatesamessagetointerestedconnectedand
unconnectedusersbyfirstselectingasubsetofconnecteduserssufficienttoreachtheunconnectedones
andsecondlybyforwardingthemessagefromtheseuserstoneighboringusersmostlikelytomeettheuncon-
nectedones.Weimplementedthishybridcommunicationmodeonasocial-basedcontext-awareubiquitous
system,SCOUT,whichwebuiltasatestingplatform.Weplantoevaluateourproposedsolutionagainstthe
existingarchitecturesbasedonthreemetricsimportanttotheperformanceofubiquitoussystems,namely:
(i) success rate or the ratio of the number of users reached to the total number of interested users, (ii) end-to-
enddelayortheaveragetimedelayincurredtoreachallinterestedunconnectedusers,and(iii)costortotal
numberofmessageforwardsrequiredtoreachalltheinterestedunconnectedusers.
21
A3 (A-Cubed)
AuthorsLakshmi Prakash (CS 2015)Sabih Bin Wasi (CS 2015)Tamim Jabban (CS 2015)
Faculty AdvisorSelma Limam Mansar, Ph.D.
CategoryInformation Systems
AbstractAstheyentertheworldofundergraduateeducation,studentsallaroundtheglobediscoveranewworld.
Universitiesusevariousmodelstoadvisethemtomakeinformeddecisionsabouttheireducation.Indeed,
atmultiplestagesduringtheiracademiclife,studentsencounteroneormoreofthefollowingproblemsin
choosing courses:
1. Itisnotalwaysobvioustoanticipatetheimpactofachangetoastudyplan,suchaspostponing,drop-
pingorchanginga‘corecourse’–especiallyinanacademicsystemwheremostcoursesareinterde-
pendent.
2. Ifvisionaroundplanningisobstructed,studentsfinditriskytoswitchmajorsinbetweentheiracademic
career.
3. Whenitcomestochoosingcourses,themerecoursedescriptionorcoursetitlemaynotbeenoughto
makegoodchoices.
A3wasinitiatedasanindependentstudyprojectinInformationSystem,andattemptstosolvesucha
demandingproblem.Basedonpatternsformajordegreeprograms,A3advisesastudentwithaplanthatis
well-fittedforastudentpursuingacareerwithvaryinginterests.Additionally,A3makesiteasiertochoose
freeelectives.A3letsusersvisualizetheirfutureundergraduateacademicplans.Itenablesstudentsto
maximizesatisfactionfromtheirundergraduatedegree,makingitmoremeaningfulfortheirprofessionallife.
Withanextensivedatabaseofcoursesofferedoncampus,A3attemptstoprovidecomprehensiveinforma-
tionabouteachcoursethatwouldaidstudentstopicktherightcoursesbeforehand.Italsoaimstobring
sociallifetotheapplicationbyintegratingpeersuggestion,rankingandadvisingfeaturesforeachcourseor
anentirecustomizedcourseplan.
A3currentlycanbeusedtoitsfullpotentialforCMU-Qstudentswhohavebeenofferedachoiceof5
majorsandvariousminors.Inthelongerrun,A¬3aimstoactasamuch-neededadvisingtoolforstudents
atanystage.Iftheycanvisualize,bendandenhancetheircourseplanwiselyandtimely,itcanunarguably
benefittheirfutureacademicandprofessionallife.
23
EZ Intern: Internship Tracking System
AuthorsHissa Al-Bahr (IS 2013)Reham Al-Tamime (IS 2013) Nabeeha Haque (IS 2013) Abhay Valiyaveettil (IS 2013)
Faculty AdvosorIan Lacey, Ph.D.
CategoryInformation Systems
AbstractThemainobjectiveofthisresearchistodesignasystemfortheOfficeofProfessionalDevelopmentinorder
toprovidestudentswithanattractiveanduserfriendlywebsitethatcontainsalltheopportunitiesforintern-
shipsandjobsavailablelocallyforCMUQstudentsandtoallowthemtoapplyonline.
Students at CMUQ receive several important and interesting emails from time to time about job/internship
opportunitiesfromtheOfficeofProfessionalDevelopmentatCMUQ.Eachofthoseemailshasadifferent
subjectandisaimedatdifferentgroupsofstudents.SometimescompaniesemailtheOfficeofProfessional
DevelopmentofCMUQtoinformstudentsaboutaspecificopportunityandthentheOfficeofProfessional
DevelopmentofCMUQforwardsittostudents.
Thebestpartaboutthissystemisthatitenablesthestudentstoapplyonlineandcanstoretheirrésumé
onlinesothattheydonothavetouploaditeachtimetheyapply.Thisideacameaboutwhenthegroup
memberswerethinkingofasystemthatcouldbeimplementedatCMU-Qandthemembersreceiveda
seriesofe-mailsregardinginternships.Thisinspiredtheteamtodevelopasystemthatcouldeliminatethe
unnecessarystackofe-mails.ThegroupgotintouchwiththeOfficeofProfessionalDevelopmentinorder
toconductsomemoreresearchandtounderstandtheprocessmoreclosely.Thisprojectisreallyimportant
sinceitcanresolveapressingissuethatCMU-Qfacesatthemoment.Withinternshipopportunitiesonthe
rise,wereallywanttohaveanorganizedsystemforstudentstoapply.
25
Lost & Found
AuthorsShivani Arora (BA 2013)Nur Aysha Anggraini (IS 2013)Fatema Akbar (IS 2013)Maryam Yousuf (IS 2013)
Faculty AdvisorIan Lacey, Ph.D.
CategoryInformation Systems
AbstractEducationCityisacommunitywithstudents,faculty,staff,andvisitorsacrosssixbranchcampuses.These
membersbelongingtodifferentcampusesinteractwitheachotheronadailybasisformeetings,cross-regis-
teredcourses.Withsomuchmovingaround,often,thesecommunitymemberstendtoloseitemswithintheir
buildingorothers.Currently,searchingfortheirlostitemsisaverytediousandtiresomeprocess–a“loser”
hastophysicallyapproachthesecuritydesksatoneormorebuildingsandaskthemaboutthelostitem.And
iftheitemisavailable,the“loser”hastofilloutseveralformstoclaimtheitem.If,ontheotherhand,theitem
hasnotbeenfoundbythesecurityguards,the“loser”mayhavetoapproachthesecuritydeskseveraltimes
againlatertofindtheiritem.
TheLost&Foundsystemautomatesthispaper-based,time-consumingprocess,makingitsimpleandconve-
nientfortheECcommunitymembers.Itservesasanonlineweb-application,whereECmemberscancheck
iftheirlostitemhasbeenfoundbyanysecuritydesksinanyoftheECbuildings,andiftheitemhasnotbeen
found,theycanreporttheirlostitem.Iftheiritemisfoundinthefuture,theywillbenotified.TheLost&Found
systemultimatelyconnectsthetwosidesofthelostandfoundprocessinanefficient,user-friendlymanner.
27
MoltaQatartans: Tartans Forum System
AuthorsWadha Alajmy (IS 2013) Mariem Fekih (IS 2013) Tasneem Jahan (IS 2013) Abhinav Vemuri (IS 2013)
Faculty AdvisorIan Lacey, Ph.D.
CategoryInformation Systems
AbstractTheposterisaboutasystemtheteamisworkingonaspartoftheJuniorProjectISclass.Theteamwaslookingforawaytolinkalumniwithcurrentstudents.AftercontactingKhadraDualehfromtheprofessionaldevelopment office and Feras Villanueva from themarketing department, the team concluded that theredoesn’texitaCMUQwebsite that joinsallmembersof thecommunity together.Since there isalreadyawebsitefortheAlumniAssociationinPittsburgh,theteamrealizedthatitwouldbeinterestingiftheQatarcampushadasimilarsystemforstudentsandalumniwhostudiedinQatar. AmongthewaysthatgraduatesstayconnectedtoCMUQare:(1)ACMUQemail,wherealumnisendemailstoKhadraorprofessors inorder tostay informedabout theupcomingevents, talks,andopportunities togiveback.(2)SocialnetworkingwebsiteslikeFacebookorTwitter,wherealumnicancommunicatewitheachotherandcurrentstudents.ThroughthesesocialwebsitestheycanstayupdatedabouttheirfriendsandalsocansharetheirexperiencesaftergraduationwithstudentsatCMUQ.
However,thesemeansremaininefficientandrepresentseveralproblems:
• ThealumnimayhavedisabledtheirCMUQemailsornolongerusethemandcheckthem. • Thealumnidon’treceiveorreplytoanyemailcomingfromCMUQorfromtheprofessional
developmentoffice.
The“TartansForumSystem”isawebsitethatwillbringinthedifferentpartsoftheCMUQcommunitytothesamepage.Itwillhelpto:
• Enhancethebondsbetweenalumniandcurrentstudentsbyenablingthemtoexchangeideas,thoughts,advice,andevenjobopportunities.
• EnablethealumnitostayintouchwithCMUQfacultyandstaff.
• KeeptheAlumniupdatedaboutupcomingeventsoncampusorevenorganizethemandinvitepeopletoattend.
29
Using Mobile Technology for Enhancing Young Qatari Health Behavior: An Experiment Design
AuthorsAysha Anggraini (IS 2013) Nawal Behih (IS 2014)Maahd Shahzad (IS 2014)
Faculty AdvisorSelma Limam Mansar, Ph.D.
CategoryInformation Systems
AbstractThis project introduces the design of an experiment that tests the effectiveness of mobile applications
foraweight lossprogram.Amobileapplicationwasdeveloped,combiningthreemodules:a tailoredtext
messagingservice,goalsettingandprogressmonitoringsupport,andasocialnetworksetup.Theintentof
theprojectistotestitseffectivenessinachievingweightmanagementgoals.Anelementoflocalizationis
addedtotheapplicationdesign,addressingtheneedsofQatar’spopulation.Whenthemobileapplication
developmentiscompletedaninterventionstudyisperformedwithparticipantsbecausetheapplicationhas
tobetested.Thisprojectdescribesausabilitystudydesign:verbalfeedbackiscollectedfromparticipants
ofthelocalcommunitywhoareinteractingwiththetechnology.Eachparticipantisaskedtocompletealist
oftasks,structuredaccordingtothethreeinterfaces.Thefeedbackallowsthedeveloperstoenhancethe
mobileapplication.ThisMoMsubmissionsummarizesprogressonanawardedUREPentitled“UsingMobile
TechnologyforEnhancingYoungQatariHealthBehavior”.
Ourcontributionhassofarbeenondevelopingsomepartsoftheapplicationanddesigningtheusabilitytesting
experiment.Wemeetweeklyasaresearchteamanddiscussprogressaccordingtoatimeline.WeusetheIS
labresources(aserver,Androidphones,Androiddevelopertools,andvoicerecorderstorecordparticipants’
feedback).ThisresearchprojectisimportantbecauseitisinlinewithQatar’sgoalstoreduceobesityand
relatedhealthproblems.Itshowsushowmoderntechnologycanbeusedtosolvehealthproblems.
31
Gettin’ the Flow; Makin’ Good Grades
AuthorsMaha Al-Moghany (CS 2012)Sara Kawas
Faculty AdvisorDavid Emmanuel Gray, Ph.D.
CategoryHumanities and Social Sciences
AbstractCsikszentmihalyi’sflowtheorydescribesastateofmindinwhichapersonbecomesfully“immersed”andengaged inanactivity.According toCsikszentmihalyi, apersoncanachieveflowevenwhendoingwhatmightbethoughtthedullestofactivities,aslongasthatperson’sskillsareonparwiththechallengeoftheactivity.Inourstudy,weexplorethisphenomenoninsidetheclassroom,determiningtheextenttowhichflowinfluencesstudentperformanceandtheirperceivedlearning.Somepreviousstudieshavefocusedonflowinlearningenvironments,butourstudyfocusesonasingleuniversity-levelcourseoveranentiresemester,andcomparesthebalanceofstudentskillsandcourseworkchallengetogrades.
StudentswhotookajointCMU-Q/NU-Qcourseinlogicalreasoning(cross-listedas80-206atCMU-QandPHIL242-70atNU-Q)weresurveyedthroughoutthesemesteraftereachquizaboutthequiz’sperceivedchallengeandabouttheirownperceivedlevelofskillinsolvingthatquiz’sproblems.Whenweanalyzedtheresults,weclusteredthestudentsintofourgroups,usingamethodproposedbyMassiminiandCarli.First,apatheticstudentsarethosewhoreportlowerthanaverageonbothchallengeandskillforaquiz.Second,anxiousstudentsarethosewhoreporthigherthanaveragechallengebut lowerthanaverageskill.Third,boredstudentsarethosewhoreportlowerthanaveragechallengebuthigherthanaverageskill.Fourthandfinally,studentsinflowarethosewhoreporthigherthanaverageonbothchallengeandskill.Attheendofthesemester,studentswereaskedtocomparetheirlogicabilitiesatthebeginningofthesemesterwithwheretheybelievedtheirabilitieswerecurrently.Wethenusedthistocalculatetheirperceivedlearningoverthesemester.
Fromouranalysis,wefoundthatstudentsinflowscoredahigheraveragegradeonquizzesthanboredorapatheticstudents.Yet,anxiousstudentsdidscoreahigheraveragegradethanstudentsinflow.However,studentsinflowreportedthehighestperceivedlearningdistanceamongallfourgroupsofstudents.Ourre-sultsindicateadifferencebetweengradesandperceivedlearningamongthefourgroups,whichfuturestud-iesoughttoexploreinordertodiscoveramoredetailedcausalrelationshipamongthesevariables.Further-more,ourresultsmayencouragefurtherworkintodiscoveringtheextenttowhichinstructorscanencourageflowintheirstudents,andwhetherthismightleadstudentstogethighergrades.
33
Service Learning at CMU-Q: Motivations, Gains, and Challenges
AuthorsShivani Arora (BA 2013)Firas Bata (BA 2013)
Faculty AdvisorSilvia Pessoa, Ph.D.
CategoryHumanities and Social Sciences
AbstractThisstudyexaminesthemotivation,gains,andchallengesexperiencedby27undergraduatestudentsat
CarnegieMellonUniversityinQatar(CMU-Q)whoparticipatedasvolunteerteachersinanadultliteracy
programformigrantworkersinQatar.Atotalof27studentreflectionswereanalyzedforrecurrentthemes,
22studentscompletedasurveyabouttheirparticipationintheprogram,andsevenstudentswereperson-
allyinterviewed.
Thefindingsindicatethatinitially,studentsinhighereducationneedincentivessuchasobtainingacademic
credittoenrollincommunityserviceprograms.Inadditiontoacademiccredit,duringtheirexperience,the
participatingstudentsgainedseveraltangibleandintangiblebenefitsincludingdevelopingseveralnew
skillssuchascommunicationandteachingskills,andanappreciationforthedifferencesinthecommunities
surroundingthem.AlthoughtheinitialmotivationforthestudentstoparticipateinLanguageBridgesmay
havebeenacademiccredit,thisinitialexperiencemaylaterleadtostudentsparticipatinginsimilarpro-
gramsinthefuturefortrulyhumanitarianandaltruistreasons.Basedonthesefindings,wediscussrecom-
mendationsforcurricularandmetacurricularpracticesatCMU-Q.
35
Challenges in Mobile Opportunistic Networks
AuthorAbderrahmen Mtibaa
Faculty AdvisorKhaled Harras, Ph.D.
CategoryPostgraduate
AbstractMobile devices such as smart-phones and tablets are becoming ubiquitous, with ever increasingcommunication capabilities. In situations where the necessary infrastructure is unavailable, costly, oroverloaded, opportunistically connecting these devices becomes a challenging area of research. Data isdisseminatedusingnodes that store-carry-and-forwardmessagesacross thenetwork. In suchnetworks,nodecooperationisfundamentalforthemessagedeliveryprocess.Therefore,thelackofnodecooperation(e.g.,anodemayrefusetoactasarelayandsettleforsendingandreceivingitsowndata)causesconsiderabledegradation in the network. In order to ensure node cooperation in such networks,we investigate threemain challenges: (i) ensuring fair resource utilization among participating mobile devices, (ii) enabling trustful communicationbetweenusers,and(iii)guaranteeingscalablesolutionsforlargenumberofdevices.
(i) Fairness isparticularly important formobileopportunisticnetworkssince itactsasamajor incentivefornodecooperation.WeproposeandevaluateFOG-areal-timedistributedframeworkthatensuresefficiency-fairnesstrade-offforusersparticipatingintheopportunisticnetwork.
(ii) Since usersmay not accept to forwardmessages in opportunistic networks without incentives, we
introduceasetoftrust-basedfilterstoprovidetheuserwithanoptionofchoosingtrustworthynodesincoordinationwithpersonalpreferences,locationpriorities,contextualinformation,orencounter-basedkeys.
(iii) Mobile opportunistic solutions should scale to large networks. Our hypothesis is that in large scalenetworks,mobile-to-mobilecommunicationhasitslimitations.WethereforeintroduceCAF,aCommunityAwareForwardingframework,whichcaneasilybeintegratedwithmoststate-of-the-artalgorithms,inordertoimprovetheirperformanceinlarge-scalenetworks.CAFusessocialinformationtobreakdownthenetworkintosub-communities,andforwardmessageswithinandacrosssub-communities.
Inthethreecontributionsweproposeabove,weadoptareal-tracedrivenapproachtostudy,analyze,andvalidate our algorithms and frameworks.Our analysis is based on differentmobility traces including theSanFranciscotaxicabtrace,tracescollectedfromconferencessuchasInfocom\’06andCoNext\’07,andDartmouthcampuswirelessdataset.
37
Characterization of Hadoop MapReduce Applications
AuthorsMohammad HammoudM. Suhail Rehman
Faculty AdvisorMajd F. Sakr, Ph.D.
CategoryPostgraduate
AbstractHadoopisanopensourceimplementationofMapReduce,oneofthemostsuccessfulrealizationsoflarge-
scaledata-intensivecloudcomputingplatforms.HadoopMapReduce isbecomingaubiquitousprogram-
mingmodelfindingapplicationsonmanyfieldsincludingweb,medicineandastronomy,amongothers.Asa
result,HadoopMapReduceisfacedwithanenormousspaceofapplicationdomainsandsystemparameters
thatmightimpactMapReduceemploymentsandstudies.Assuch,characterizingHadoopbecomescrucial
for:(1)fosteringtheunderstandingoftheMapReducemodelsoastoenablepracticalapplicationsofMapRe-
duceprograms,(2)gleaninginsights intotheframework’sbottlenecksandpotentially identifyingdesirable
optimizations,(3)establishingaquantitativefoundationforevaluatingMapReduceapplications,and(4)sup-
portingresearchandcomparisonsacrossmultiplerelatedstudies.Weproposeandapplyacharacterization
methodologythatservesinmeetingtheseobjectives.
TheonlytwocurrentstudiesoncharacterizingHadoopfocusprimarilyonpromotingbenchmarksuitesand
overlookestablishingamethodologythroughwhichHadoopcanbeeffectivelycharacterized. Incontrast,
wefocusondevelopinganeffectivecharacterizationmethodologyandsuggestusingthatmethodologyata
laterstagetocreateastandardbenchmarksuite.WeutilizetworepresentativeandwidelyadoptedMapRe-
ducebenchmarksandcharacterizeHadoopalongsevendimensions:thedegreeofparallelism,thedataset
distribution,thejobsystemresourceutilization,thejobcommunicationtocomputationratio,theexecution
timelines,thedatasettypesandsizes,andthejobnetworktrafficpattern.Asaresult,weattainedmultiple
observations.Forinstance,werealizedthathavingthenumberofmappersequaltothenumberofavailable
mapslotsleadstothebestparallelism.Thisdoesnotonlyimproveperformance,butalsotacklesoneofthe
rootcausesofthecommonstragglersprobleminMapReduce.Asanotherexample,weobservedthatthe
networkbandwidthdissipationofthereducestageexceedsthatoftheshufflestagebyapproximatelyafac-
toroftwo.Thismightsuggestaddressingthenetworkbandwidthproblematthereducestage,especially
sinceallresearchefforthasbeenputsofarontheshufflestage.
39
CoGRS: A Center-of-Gravity Reduce Task Schedule for MapReduce
AuthorsMohammad HammoudM. Suhail Rehman
Faculty AdvisorMajd F. Sakr, Ph.D.
CategoryPostgraduate
Abstract:MapReduceisoneofthemostsuccessfulrealizationsoflarge-scaledata-intensivecloudcomputingplat-
forms.Ascomparedtotraditionalprogrammingmodels,MapReduceautomaticallyandefficientlyparallelizes
computationbyrunningmultipleMapand/orReducetasksoverdistributeddataacrossmultiplemachines.
Hadoop,anopensourceimplementationofMapReduce,schedulesMaptasksinthevicinityoftheirinput
splitstherebyreducingnetworktraffic.However,forreducetasks,Hadoopneitherexploitsdatalocalitynor
addressesdatapartitioningskewinherentinmanyMapReduceapplications.Consequently,MapReduceex-
periencesaperformancepenaltyandnetworkcongestionasobservedinourexperimentalresults.
Inthisstudy,weintroducetheCenter-of-GravityReducetaskScheduler(CoGRS),apracticalstrategyforim-
provingMapReduceperformanceinclouds.CoGRSattemptstoscheduleeachreduceratitscenter-of-grav-
itynode.Itcontrollablyavoidsschedulingskew,asituationwheresomenodesreceivemorereducersthan
others,andpromoteseffectivepseudo-asynchronousMapandReducephasesresultinginearliercompletion
ofsubmittedjobs,diminishednetworktraffic,andbetterclusterutilization.
WeimplementedCoGRSinHadoop-0.20.2andconductedextensiveexperimentationstoevaluateitspoten-
tial.WefoundthatCoGRSoutperformsthenativeHadoopschedulerby11%,andbyupto26%inruntime
performanceforourbenchmarkstudies.Inaddition,wedeployedCoGRSonAmazonEC2andfoundthat
ourstrategyscaleswell,reducingnetworktrafficbyasmuchas38.6%,whichconsequentlytranslatesinto
anruntimeimprovementofupto23.8%.Fromthesestudies,webelievethatCoGRSisapplicabletoseveral
cloud computing environments and applications, including but not limited to, shared environments and sci-
entificapplications.LASARpavesthewayfortheseapplications,andothers,tobeportedtovariousclouds
inaneffectivemanner.
41
GreenLoc: Energy Efficient Wi-Fi-based Indoor Localization
AuthorMohamed Abdellatif
Faculty AdvisorKhaled Harras, Ph.D.
CategoryPostgraduate
AbstractUser-localizationandpositioningsystemshavebeenacorechallengeinthedomainofcontext-aware
pervasivesystemsandapplications.GPShasbeenthede-factostandardforoutdoorlocalization;however,
geo-satellitesignalsuponwhichGPSrelies,areinaccurateinindoorenvironments.Therefore,variousin-
doorlocalizationtechniquesbasedontriangulation,sceneanalysis,orproximity,havebeenintroduced.The
mostprominenttechnologiesoverwhichthesetechniquesareappliedincludeWiFi,Bluetooth,RFID,Infra-
red,andUWB.Duetotheubiquitousdeploymentofaccesspoints,WiFi-basedlocalizationviatriangulation
hasemergedtobeamongthemostprominentindoorpositioningsolutions.Amajordeploymentobstacle
forsuchsystems,however,isthehighenergyconsumptionratesofWiFiadaptersinmobiledeviceswhere
energyisthemostvaluableresource.
WeproposeGreenLoc,anindoorgreenlocalizationsystemthatexploitssensorsprevalentintoday’ssmart-
phonesinordertodynamicallyadaptthefrequencyoflocationupdatesrequired.Significantenergygains
can,therefore,beacquiredwhenusersarenotmobile.Forexample,accelerometerscanaidindetect-
ingdifferentuserstatessuchaswalking,runningorstopping.Basedonthesestates,mobiledevicescan
dynamicallydecideupontheappropriateupdatefrequency.Weaccommodatevariousmotionspeedsby
estimatingthevelocityofthedeviceusingthelatesttwolocationcoordinates,andthetimeintervalbetween
thesetworecordedlocations.WehavetakenthefirststepstowardsimplementingGreenLoc,basedon
theinfamousEkahausystem.Wehavealsoconductedpreliminarytestsutilizingtheaccelerometer,gravity,
gyroscope,andlightsensorsresidingontheHTCNexusOneandIPhone4smart-phones.
Tofurthersaveenergyintypicalindoorenvironments,suchasmalls,schools,andairports,GreenLoc
exploitspeople’sproximitywhenmovingingroups.Deviceswithinshort-rangeofeachotherdonotneces-
sarilyrequirethattheyeachbeindividuallytracked.Therefore,GreenLocdetectsandclustersusersmoving
togetherandelectsareferencenode(RN)basedondeviceenergylevelsandneeds.TheelectedRNwill
thenbetrackedviatriangulationwhileothernodesinthegroupwillbetrackedbasedontheRN’slocation
usingBluetooth.Ourinitialanalysisdemonstratesverypromisingresultswiththissystem.
43
Hala 2.0: Considerations for Developing a Test Bed for Multi-Lingual, Cross-Cultural Human Robot Interaction
AuthorsMicheline ZiadeeImran FanaswalaMaxim MakatchevAmna Al ZeyaraHuda GedawyNawal Behih
Faculty AdvisorMajd Sakr, Ph.D.Reid Simons, Ph.D.
CategoryPostgraduate
AbstractThis research focuses on developingmulti-lingual, cross-cultural human robot interaction by identifying,analyzing, and utilizing language and culture-related factors/variables that influence human interactions.Ourtestbedisabi-lingual,cross-culturalrobotreceptionist,Hala,deployedattheCarnegieMellonQatarreceptionthathelpsuscollectthenecessarydata.HalacanspeakArabicandEnglishandsheinteractswithavarietyofusers.
Halaasaplatformmodelsuserinteractions,invitesuserswithaSICKlaser,handlesfacialanimations(borrowedfromPaulEkman’smodels), text-to-speechand lipsynchronization (borrowedEnglishvisemes forArabicspeech),aswellaserrorandreporting,postdialogueanalysis,networking/interprocesscommunication,andarichclientinterface.
Resultsfromourpriorworkindicatedvariationsinuserinteractionswiththerobot.Forexample,around97%oftheinteractionswithHalawereinEnglishbuttheaveragedurationwashigherforinteractionsinArabicasopposedtointeractionsinEnglish(231.7secondsforArabic/184.4secondsforEnglish).Also,ArabicnativespeakersweretwiceaslikelytoacceptaninvitefromtherobotthanEnglishnativespeakers.UsersinQatar,ascomparedtothoseintheUS,thanktherobotlessfrequently.
In this work, we discuss Hala 2.0, which has more Arabic features in expression and interaction. WeconstructedHala’spersonalitytakingintoaccountthesocio-culturalcontextinwhichherinteractionswilltakeplace.WehaveexpandedHala’sstatelesscontent,inEnglishandinArabic,toincludepreviouslyunansweredqueriesthatwegatheredfromHala’slogs.Wehaveintroducednaturallanguageprocessingwithsemanticandsyntacticparsingandcreatednewstate-fullcontentwherebyHalaiscapableofcarryingonameaningfulconversation. WedevelopedArabicvisemestoreplacetheborrowedEnglishvisemesthatwereusedforArabicspeech.Inadditiontothat,wedevelopedmorefacialanimationsthataddtoHala’sabilitytoexpressemotions;ourlateststudyfocusedonunderstandingthedifferencesinaccentsoffacialexpressionsthatvarybetweencultures–inthiscase,ArabicandAmerican.Inanotherstudy,weinvestigatedthestrategiesusedbyEnglishandArabicnativespeakerswhengivingdirections.ResultsfromthesestudieswillinfluencethedesignofHala2.0.
45
Performance Prediction of MapReduce Applications in Elastic Compute Clouds
AuthorFan Zhang
Faculty AdvisorMajd F. Sakr, Ph.D.
CategoryPostgraduate
AbstractTheMapReduceprogrammingmodelisawidelyacceptedsolutiontoaddresstherapidgrowthoftheso-
calledbig-dataprocessingdemands.Variousapplicationswithavery largevolumeof inputdatacan run
on an elastic compute cloud composed ofmany distributed computing instances. This elastic compute
cloudisbestrepresentedbyavirtualcluster,suchasAmazonEC2.PerformancepredictionofMapReduce
applicationsischallengingduetothecomplexinteractionoftheMapReduceframeworkrunningonhighly
parameterizeddistributedvirtualizedresources.
Inthisstudy,wehavecharacterizedaseriesofMapReduceapplicationsonAmazonEC2,andidentifiedhow
inputdatasizeandclustersizeaffecttheexecutiontime.Westudiedthescalingcurveofeachapplication,
andprovidedanumberofdatacleaning,miningandpredictionmethodstodiscoverthedatapattern.These
MapReduceapplicationsspanacrossdata-intensive, compute-intensive, and iterativebenchmarks. Initial
observationssuggestanearpositivepower-lawdistributionofexecutiontimeagainsttheinputdatasizeand
anegativepower-lawdistributionagainsttheclustersize.
Basedonalltheseobservations,wehaveestimatedthepredictionerrorandsuggestedmethodstoreduce
the estimation error. Five regression and predictionmethods, linearly incremental regression, polynomial
regression, exponential regression, moving average regression and power regression are thoroughly
investigatedandcomparedforestimatingpredictionerrors.Weconcludethatthepowerregressionperforms
bestatperformancepredictioncomparedwiththeothermethodsevaluated.
Our observations andperformancepredictionmethodswill aid users in choosing appropriate computing
resources, both virtual and physical, from small-scale experimental test runs. This approachwill predict
performancespeedupsor slowdowns forMapReduceapplicationswhenscaling the infrastructureor the
inputdatasets.
47
SmartReader: A natural language processing-based active and interactive system for accessing English language content and advanced language learning
AuthorsKemal OflazerTeruko MitamuraHideki ShimaJun ArakiAhmed Salama
Faculty AdvisorKemal Oflazer, Ph.D
CategoryPostgraduate
Abstract:SmartReader is ageneral-purpose “readingappliance”being implementedatCarnegieMellonUniversity(QatarandPittsburgh)–buildinguponanearlierprototypeversion.Itisanartificialintelligencesystemthatemploysadvancedlanguageprocessingtechnologiesandcaninteractwiththereaderandrespondtoque-riesabout thecontent,wordsandsentences inatext.Weexpect it tobeusedbystudents inQatarandelsewheretohelpimprovetheircomprehensionofEnglishtext.
SmartReader ismotivatedby theobservation that text isstill thepredominantmediumfor learningespe-ciallyattheadvancedlevelandthattext,being``bland,’’ishardlyaconduciveandmotivatingmediumforlearning.Thisisespeciallytruewhenonedoesnothaveaccesstotoolsthatenableonegetoverlanguageroadblocks,rangingfromunknownwordstounrecognizedandforgottennames,tohard-to-understandsen-tences.SmartReaderstrivestomakereading(English)textualmaterial,an“active”andan“interactive”pro-cesswiththeuserinteractingwiththetextusinganytime-anywherecontextually-guidedquerymechanismbased-oncontextualuserintentrecognition.WithSmartReader,ausercan:
• Inquireaboutthecontextuallycorrectmeaningorsynonymsofawordoridiomaticandmulti-wordcon-structions.
• Selectaperson’sname,andthengetanimmediate `flashback’’tothefirst(orthelast)timethepersonwasencounteredintexttoremindherselfthedetailsoftheperson.
• Extractasummaryofasectiontorememberimportantaspectsofthecontentatthepointsheleftoff,andcontinuereadingwithasignificantlyrefreshedcontext.
• SelectasentencethatshemaynotbeabletounderstandfullyandaskSmartReadertobreakitdown,simplifyorparaphrasetocomprehenditbetter.
• Testhercomprehensionofthetextinapageorachapter,byaskingSmartReadertodynamicallygeneratequizzesandansweringthem.
• Askquestionsaboutthecontentofthetextandgetanswersinadditiontomanyotherfunctions.
SmartReaderisbeingimplementedasamulti-platform(tablet/PC)client-serversystemusingHTML5tech-nology,withUnstructuredInformationManagementArchitecture–UIMAtechnology(usedrecentlyinIBM’sWatsonQ/AsystemintheJeopardyChallenge)astheunderlyinglanguageprocessingframework.
49
VOtus: A Flexible and Scalable Monitoring Framework for Virtualized Clusters
AuthorsSuhail RehmanMohammad Hammoud
Faculty AdvisorMajd Sakr, Ph.D.
CategoryPostgraduate
AbstractCloudcomputingrevolutionizesthewaybigdataisprocessedandoffersacompellingparadigmtoorga-
nizations.Data-intensivescientificapplicationsarebeingportedtocloudenvironmentssuchasvirtualized
clusters;however,thisprocessposesitsownsetofchallenges.
Given thecomplexityof theapplicationexecutionenvironmentaswellas the infrastructure, routine tasks
suchasmonitoring,performanceanalysisanddebuggingofapplicationsdeployedon thecloudbecome
cumbersomeandcomplex.Thesetasksoftenrequirecloseinteractionandinspectionofmultiplelayersin
theapplicationsoftwarestack.Forexample,whenanalyzingadistributedapplicationthathasbeenprovi-
sionedonaclusterofvirtualmachines,aresearchermightneedtolookatthevirtualresourceusage(e.g.,
virtualCPUandvirtualmemory)andthecorrespondingphysicalresourceusage(physicalCPUandphysical
memory)ofthecluster.Thiswouldrequiretwodifferentsetsoftoolstocollectandanalyzeperformancemet-
ricsfromeachlevel.OnesuchtoolisOtus,whichcurrentlyreportsonlythevirtualresourceutilizationandnot
thephysicalresourceutilizationonvirtualizedclusters.
ThroughVOtus,wehaveextendedOtustoincludephysicalresourcemetricsthatcanbecollectedfromthe
hypervisor.Forexample,aresearchercannowviewthevirtualresourceusageofhisapplicationontheVMs
aswellasthephysicalresourceusageoftheVMsonthephysicalmachines.Thisenablestheresearcherto
closelymonitortheapplicationandmakemodificationsattheapplicationlevelorattheVMlevel.Thiswould
helptheresearchertooptimizeperformanceormanageinfrastructureeffectively.VOtusalsoscalestolarge
clustersandcanbeusedforreal-timemonitoringandtoarchiveperformancemetricsfordetailedanalysis
inthefuture.VOtusshouldprovetobeanimportanttoolforresearcherswhoplantodesign,developand
deploydistributedapplicationsonvirtualizedclusters.
51
Formorethanacentury,CarnegieMellonUniversityhasbeeninspiring
innovationsthatchangetheworld.Consistentlytopranked,Carnegie
Mellonhasmorethan11,000students,90,000alumniand5,000faculty
andstaffglobally.
In 2004, Qatar Foundation invited Carnegie Mellon to join Education
City,agroundbreakingcenter forscholarshipandresearch.Students
from 39 different countries enroll at our world-class facilities in
EducationCity.
Carnegie Mellon Qatar offers undergraduate programs in biological
sciences, business administration, computational biology, computer
scienceandinformationsystems.
Learnmoreatwww.qatar.cmu.edu
P. O . B o x 2 4 8 6 6 | E d u c a t i o n C i t y, D o h a , Q a t a r | P h : + 9 7 4 4 4 5 4 8 4 0 0 | F a x : + 9 7 4 4 4 5 4 8 4 1 0 | w w w. q a t a r. c m u . e d ux 2 4 8 6 6 | E d u cP. O . B o x Q a t a r | P h : + 9QQn C i t y, D o h a , QQQ x : + 9 7 4 4 4 5 4 84 5 4 8 4 0 0 | F a x m u . e d u| w w w. q a t a r. c