Situation Recognition Using EventShop

152 Pages·2016·4.672 MB·English

by  Vivek K. Singh, Ramesh Jain (auth.)

Vivek K. Singh · Ramesh Jain
Situation
Recognition
Using EventShop
Situation Recognition Using EventShop

Vivek K. Singh • Ramesh Jain
Situation Recognition Using
EventShop
123
VivekK.Singh RameshJain
RutgersUniversity UniversityofCalifornia
NewBrunswick,NJ,USA Irvine,CA,USA
ISBN978-3-319-30535-6 ISBN978-3-319-30537-0 (eBook)
DOI10.1007/978-3-319-30537-0
LibraryofCongressControlNumber:2016936667
©TheAuthor(s)2016
Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartof
thematerialisconcerned,specificallytherightsoftranslation,reprinting,reuseofillustrations,recitation,
broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,andtransmissionorinformation
storageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilarmethodology
nowknownorhereafterdeveloped.
Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication
doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant
protectivelawsandregulationsandthereforefreeforgeneraluse.
Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthisbook
arebelievedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsor
theeditorsgiveawarranty,expressorimplied,withrespecttothematerialcontainedhereinorforany
errorsoromissionsthatmayhavebeenmade.
Printedonacid-freepaper
ThisSpringerimprintispublishedbySpringerNature
TheregisteredcompanyisSpringerInternationalPublishingAGSwitzerland
VKS:
To mygreat-grandfather,ShriParsadiLal.
RJ:
To allmyresearch partners.

Preface
Imagine a scenario where millions of aspects of the physical world around us,
including weather patterns, stock prices, social phenomena, traffic information,
disease incidents, and so on, are accessible to anybodywho so desires. Similarly,
eachindividualhasaccesstorichpersonaldataaboutthemselvesrangingfromtheir
heartratestomovementpatterns,totemperature,togaze,tosocialinteractions,and
soon.Suchascenarioisnotmerelysciencefiction,butratherincreasinglybecoming
arealityformorepeoplethaneverbeforeatapacefasterthaneverseenbefore.
This marks a significant change in the human thinking about data. In the past,
ourscientificthinkinghasoftenbeencloudedbythescarcityofdata;wecollected
data only when really needed and the potential insights to be drawn from it were
defined even before the data was created. Increasingly we have an abundance of
data,i.e.,wehavemoredatathanwhatwehavethetoolsandabilitiestocomprehend
and convert into actionable insights. This means that the bottleneck for making
significantprogressinfieldsrangingfromhealthtobusinessanalysisandtodisaster
response is no longer the existence of data but rather the tools and techniques to
makesenseofthedata.
This book makes a contribution toward converting these multitudes of data
streamsinto actionableinsights. Specifically it computationallydefinesthe notion
of situations as an abstraction of millions of data points into actionable insights
for individuals, describes a computationalframework to model and evaluate such
situations, and presents an open-source web-based system called EventShop that
canbeusedbydifferentuserstocreatetheirownsituation-awareapplications.
Weparsethesegoalsintothreedifferentpartsofthebook.Thefirstpartfocuses
onUnderstandingandDefiningSituations,the second oneproposesa Framework
forRecognizingSituations,andthethirdonedescribesanoperationalizationofthis
framework as EventShop: An Open-Source Toolkit for Situation Recognition. We
concludethebookwith a summaryofthecontributionsmadeaswellasthemany
remainingchallengesanddirectionsforfuturework.
PartIonUnderstandingandDefiningSituationsincludesthreechapters,which
focus on painting a vision for the data-rich eco-system that we are stepping into
andhowsituationrecognitionwillplayanimportantroleinsuchasetting.Thenext
vii
viii Preface
chapter reviews the interpretations of situations as used in different domains and
synthesizes them into a computational definition for situations. The third chapter
reviewsrelatedworkwhichcoversdifferentaspectsofthisworkincludingdefining
situations,buildingcomputationalmodelsfortheirrecognitionfromheterogeneous
data,andcreatingtoolkitsthatallowusersto experimentwith andbuildtheirown
situationrecognitionapplications.
Part II describes a Framework for Recognizing Situations and includes three
chapters. The first chapter discusses the overall framework including the process
of situation modeling, situation recognition, and personalized alerts. The next
chapterzoomsintotheprocessofcreatingS2S(situation-to-source)models,which
are schematic blueprints for the conversion of the high level, and often vague,
mental models of situation of interest (e.g., epidemic) into detailed list of data
sources and operators needed to recognize them. The third chapter describes the
computational framework needed to implement and test the S2S models. This
includes the definition of appropriate data structures and an algebra of situation
recognitionoperators,whichcanbecombinedtogethertodefinearbitrarilycomplex
situationrecognitionmodules.
PartIIIfocusesonEventShop:An Open-SourceToolkitforSituationRecogni-
tion.Itincludesfourchaptersandstartswithadescriptionofthesystemarchitecture
of EventShop. The next chapter focuses on the practical usage of EventShop
and discusses how different users can configure and build their own situation
recognitionapplicationsusingEventShop.Thethirdchapterhighlightscasestudies
onusingEventShopforpracticalapplicationsincludingwildfirerecognitionusing
satellite imagery and social data, flood evacuation in Thailand, and personalized
asthma/allergyrecommendationsystem.Thelastchapterprovidesasummaryofthe
contributionsofthisbookanddiscussesthemanyopenchallengesinthisimportant
researcharea.
Weexpectthisbooktoservetwoprimarypurposes.One,itactsasaprimerfora
dataenthusiastoraninformationprofessionalinterestedinharnessingthevalueof
heterogeneous,“big”,dataforbuildingdiversesituation-basedapplications.Second,
itcouldbeusedasareferencetextbyresearchersworkingintheareasrangingfrom
databasedesigntomultimodalconceptrecognition,tomiddle-wareandubiquitous
computingto design and developframeworksthat allow users to create their own
situationrecognitionapplications.
Infact,inkeepingwiththesegoals,weconsiderthelastchapteronopen-research
questionsacriticalcomponentofthebook.Beingambitiousinitsgoals,thebook’s
contributionslieasmuchindefiningtheproblemstatementsastheydoinproviding
the solutions. Clearly, many of the solutions or approaches proposed are what
should be termed as early efforts. There exist multiple opportunities to enhance
and improve the proposed framework to tackle relevant challenges including
scalability, user experience, and privacy. Based on early application experience,
we are confident that a framework so developed would have the potential to
transformmultipleaspectsofhumanlifeincludingtraffic,health,businessanalysis,
politicalcampaignmanagement,cybersecuritymonitoring,disasterresponse,crisis
mitigation,andhomelandsecurity.
Preface ix
It is with this hope of sharing our excitement, and urging you to join in this
efforttoward democratizingthe process of generatingactionable insights from all
theworld’sdata,thatwepresentyou,ourdearreaders,thisbook.
Yourssincerely,
NewBrunswick,NJ,USA VivekK.Singh
Irvine,CA,USA RameshJain
December2015