Sentic Computing: Techniques, Tools, and Applications

165 Pages·2012·2.985 MB·English

by  Erik Cambria, Amir Hussain (auth.)

SpringerBriefs in Cognitive Computation
For furthervolumes:
http://www.springer.com/series/10374
Erik Cambria Amir Hussain
•
Sentic Computing
Techniques, Tools, and Applications
123
ErikCambria AmirHussain
Media Laboratory Department of ComputingScience
Massachusetts Instituteof Technology Universityof Stirling
Cambridge Stirling
USA UK
ISSN 2212-6023 ISSN 2212-6031 (electronic)
ISBN 978-94-007-5069-2 ISBN 978-94-007-5070-8 (eBook)
DOI 10.1007/978-94-007-5070-8
SpringerDordrechtHeidelbergNewYorkLondon
LibraryofCongressControlNumber:2012942029
(cid:2)TheAuthor(s)2012
Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartof
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation,broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,andtransmissionor
informationstorageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilar
methodology now known or hereafter developed. Exempted from this legal reservation are brief
excerpts in connection with reviews or scholarly analysis or material supplied specifically for the
purposeofbeingenteredandexecutedonacomputersystem,forexclusiveusebythepurchaserofthe
work. Duplication of this publication or parts thereof is permitted only under the provisions of
theCopyrightLawofthePublisher’slocation,initscurrentversion,andpermissionforusemustalways
beobtainedfromSpringer.PermissionsforusemaybeobtainedthroughRightsLinkattheCopyright
ClearanceCenter.ViolationsareliabletoprosecutionundertherespectiveCopyrightLaw.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publicationdoesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexempt
fromtherelevantprotectivelawsandregulationsandthereforefreeforgeneraluse.
While the advice and information in this book are believed to be true and accurate at the date of
publication,neithertheauthorsnortheeditorsnorthepublishercanacceptanylegalresponsibilityfor
anyerrorsoromissionsthatmaybemade.Thepublishermakesnowarranty,expressorimplied,with
respecttothematerialcontainedherein.
Printedonacid-freepaper
SpringerispartofSpringerScience+BusinessMedia(www.springer.com)
To Laura,
For believing in me and fostering my
passions and dreams, always
Foreword
When the authors asked me to write the foreword to this book, I readily accepted
sinceitisasubjectthatintriguesmeandoneforwhichIrealizedthatIhavemuch
to learn. It was only as I got deeper into this work that I realized what a seminal
andimportantcontributiontoinformationtechnologythisbookhasthepotentialto
be.ThisbookwilllikelybecomeaclassicinthefieldofSenticComputingandbe
adopted as a textbook for students of the subject. It provides not only a compre-
hensive overview of the foundational elements of the topic, but introduces new
methods along with teaching and research challenges for the future.
There is no doubt of the impact that the World Wide Web has had on human
communication and discourse and will continue to have in the unforeseeable
future.Asitscommunicationscapabilitiesgrow,itsskillsmustalsogrow.Idonot
meantheskillsofWebusers,buttheskillsoftheWebitself.ThefutureWebwill
not just tie together documents, data, and people, it will also become a vastly
connected world of processes, objects, and things all of which will have unique
identitiesandpersonalities.LikethehumanswhousetheWeb,theseobjectsmust
have common sense, be affective, and perhaps even emotional. To be effective,
theymustbeabletomakedecisionsandjudgementssomeofwhicharelikelytobe
quitesubtle.Suchqualitiesoftheseobjectswillbeembedded,hidden,anddefined
within the data used by these entities. This data will become the DNA that
prescribesthebehavioroftheseobjects.Likeitsbiologicalcounterpart,thisDNA
willcontain theprogramming instructions for data manipulationgoverning object
actions. But what will be the source of this data?
Web technology wasrevolutionized whenusersdiscoveredthe abilitytouse it
as a platform for expressing opinions and evaluations. The previously static Web
became a dynamic corpus of user-generated informationsuch asproduct reviews,
statisticalpollingdata,surveydata,etc.Withinthisinformation,justwaitingtobe
uncovered,isawealthofconceptual,expressive,andemotionalcontentreflecting
user responses to natural language questions and options. Unfortunately, the
majorityofthiscontentisunstructuredtextcontainingalloftheambiguitiesfound
inspokencommunications.Thechallengebecamehowtoconvertthisinformation
into structured data formats required for efficient and accurate computer analysis.
vii
viii Foreword
It was soon realized that intelligent mining of this data was a difficult task
requiring the definition of new paradigms and the development of innovative
analysistools.Forexample,isitpossibletoperformclause-levelsemanticanalysis
of essentially freeform textual data which accurately allows the inference of both
conceptual and emotional information? Employing methods from artificial intel-
ligence, data mining, and such diverse disciplines as behavioral and sentiment
analysis, the authors of this book explain why the answer is yes. The result is
Sentic Computing.
Potential applications of Sentic Computing are widespread. Analyzed data
drawn from product ratings and reviews are a marketer’s dream and provide the
newcurrencyfore-commercewebsitesandsocialmedia.Theauthorsdescribehow
Sentic Computing techniques can influence HCI (Human-Computer Interaction)
andEHealth.Readersofthebookwilllikelyrealizeotherpossibleapplicationsof
the methods andtoolsina range of otherdisciplines.
I encourage the reader to carefully study this book because doing so will
provide a stimulating introduction to an area of study with great future potential.
Itwillchangeyourmindaboutthevalue(andthepotentialusage)oftheopinions
and evaluations that you make everyday. It did so for me.
Stanford, 9th April 2012 Bebo White
Preface
The ways people express their opinions and sentiments have radically changed in
thepastfewyearsthankstotheadventofsocialnetworks,Webcommunities,blogs,
wikis,andotheronlinecollaborativemedia.Thedistillationofknowledgefromthis
huge amount of unstructured information can be a key factor for marketers who
wanttocreateanimageoridentityinthemindsoftheircustomersfortheirproduct,
brand,ororganisation.Theseonlinesocialdata,however,remainhardlyaccessible
tocomputers,astheyarespecificallymeantforhumanconsumption.Theautomatic
analysis of online opinions, in fact, involves a deep understanding of natural
language text bymachines, fromwhichwe are still very far.
Hitherto, online information retrieval has been mainly based on algorithms
relyingonthetextualrepresentationofWebpages.Suchalgorithmsareverygood
at retrieving texts, splitting them into parts, checking the spelling, and counting
theirwords.Butwhenitcomestointerpretingsentencesandextractingmeaningful
information,theircapabilitiesareknowntobeverylimited.Existingapproachesto
opinion mining and sentiment analysis, in particular, can be grouped into three
maincategories:keywordspotting,inwhichtextisclassifiedintocategoriesbased
onthepresenceoffairlyunambiguousaffectwords;lexicalaffinity,whichassigns
arbitrarywordsaprobabilisticaffinityforaparticularemotion;statisticalmethods,
which calculate the valence of affective keywords and word co-occurrence
frequencies on the base of a large training corpus.
Earlyworksaimedtoclassifyentiredocumentsascontainingoverallpositiveor
negativepolarity,orratingscoresofreviews.Suchsystemsweremainlybasedon
supervised approaches relying on manually labeled samples, such as movie or
product reviews where the opinionist’s overall positive or negative attitude was
explicitly indicated. However, opinions and sentiments do not occur only at
document level, nor are they limited to a single valence or target. Contrary or
complementary attitudes toward the same topic or multiple topics can be present
acrossthespanofadocument.Inmorerecentworks,textanalysisgranularityhas
beentakendowntosegmentandsentencelevel,e.g.,byusingpresenceofopinion-
bearing lexical items (single words or n-grams) to detect subjective sentences,
or by exploiting association rule mining for a feature-based analysis of product
ix
x Preface
reviews. These approaches, however, are still far from being able to infer the
cognitive and affective information associated with natural language as they
mainlyrelyonknowledgebasesthatarestilltoolimitedtoefficientlyprocesstext
at sentence level. In this book, common sense computing techniques are further
developed and applied to bridge the semantic gap between word-level natural
languagedataandtheconcept-levelopinionsconveyedbythese.Inparticular,the
ensemble application of graph mining and multi-dimensionality reduction tech-
niques on two common sense knowledge bases was exploited to develop a novel
intelligent engine for opendomain opinion mining and sentiment analysis. The
proposed approach, termed sentic computing, performs a clause-level semantic
analysis of text, which allows the inference of both the conceptual and emotional
informationassociatedwithnaturallanguageopinionsand,hence,amoreefficient
passage from (unstructured) textual information to (structured) machine-process-
able data.
The engine was tested on three different resources, namely a Twitter hashtag
repository, a LiveJournal database, and a PatientOpinion dataset, and its perfor-
mance compared both with results obtained using standard sentiment analysis
techniquesandusingdifferentstate-of-the-artknowledgebasessuchasPrinceton’s
WordNet, MIT’s ConceptNet, and Microsoft’s Probase. Differently from most
currentlyavailableopinionminingservices,thedevelopedenginedoesnotbaseits
analysis on a limited set of affect words and their co-occurrence frequencies, but
rather on common sense concepts and the cognitive and affective valence con-
veyedbythese.Thisallowstheenginetobedomain-independentand,hence,tobe
embedded in any opinion mining system for the development of intelligent
applications in multiple fields such as Social Web, HCI, and e-health. Looking
ahead,thecombinednoveluseofdifferentknowledgebasesandofcommonsense
reasoning techniques for opinion mining proposed in this work, will, eventually,
pave the way for development of more bio-inspired approaches to the design of
natural language processing systems capable of handling knowledge, retrieving it
when necessary, making analogies, and learning from experience.
Acknowledgments
Thecompletionofthisbookwouldnothavebeenpossiblewithoutthecontribution
of many individuals, to whom I express my appreciation and gratitude. Because
thecontentsofthisbookaremainlydrawn frommydoctoral researchwork,Iam
first of all deeply indebted to my Ph.D. supervisors, namely: Amir Hussain, who
guided me every step of the way and was an immense source of inspiration
throughout;CatherineHavasi,whoseguidance,encouragement,andsupportinthe
past three years has been simply invaluable; and Chris Eckl, who helped me to
look at my research from many different points of view.
I am also grateful to all the other mentors and colleagues who supported me
duringmyscientificmissionsandinternships,inparticular,RobertSpeer,Kenneth
Arnold, Dustin Smith, Jason Alonso, and Henry Lieberman, who assisted me in
gettingstarted,using,andfurtherdevelopingcommonsensecomputingtools(both
during my research visits at MIT Media Lab and remotely); Campbell Grant, for
his critical opinions and sound advice about the commercial aspects of my
researchwork;AnnaEsposito,forherunwaveringsupportwithintheCOST2102
programme; Thomas Mazzocco and Marco Grassi, for their invaluable research
contributions; Evan Magill and Kang Li, for their insights and assistance in
refining the contents of this book; Joseph Lyons and Carmen Tropeano, for their
helpindesigningsomeofthegraphics;andJamesMunro,forthesupportandthe
data provided within my work in the field of patient opinion mining.
Special thanks also go to Praphul Chandra and Sudhir Dixit, who helped me
expandthehorizonsofmyresearchduringmyinternshipatHewlett-PackardLabs
India; Tariq Durrani, Cheng-Lin Liu, Tieniu Tan, and Derong Liu, for their gui-
dance during my research visit at the National Laboratory of Pattern Recognition
(NLPR) in the Institute of Automation of the Chinese Academy of Sciences; and
HaixunWangandYangqiuSong,forhelpingmeimprovemyskillsandexpertise
in the field of knowledge-based systems during my internship at Microsoft
Research Asia. A last, but not least, acknowledgement goes to my family and all
the old and new friends who, in the past three years, have cheered me up in
difficultmoments,celebratedwithmeformyachievements,andneverblamedme
for being often far away from them.
xi
xii Acknowledgments
My Ph.D. was one of the best experiences of my life as it gave me the possi-
bilityto:workshouldertoshoulderwithscientistsfromtopresearchinstitutes,get
in touch with both far West and far East cultures, meet special people who will
always be part of my life, and even exchange ideas with beautiful minds such as
TimBerners-Lee,theinventoroftheWeb,BeboWhite,WebpioneerintheUnited
States, and Marvin Minsky, one of the fathers of AI.
Boston, 9th May 2012 Erik Cambria