Applications of Games With A Purpose

CGAMES 2013
The 18th International Conference on Computer Games
Applications of Games With A Purpose
Suhas Aggarwal
[email protected]
Abstract—In this paper, we describe a gaming portal which
host different games with a purpose. Games include Similar
choices which help in E-commerce analytics, Label Graphs which
help in comprehensive graphs analysis, Tag Earth for Geo-Data
Tagging, Game for tagging complex objects such as news,
articles, art. It also features a fun game ‘Pick Quality’ based on
ESP game architecture which is for fun/entertainment. Portal
also has other features such as classic games, online shopping
where credits obtained from playing games can be utilized.
Keywords-Games with a purpose, Ecommerce Analytics, Graph
Analytics, Tagging articles, videos, art, Tagging Geo-Data, Gaming
Portal.
I.
INTRODUCTION
Games with a purpose utilize collective intelligence, which
is described as combining “behavior, preferences, or ideas of a
group of people to create useful results” [1]. Collective
intelligence from groups of people often produces better
results than individual subject specialists [2]. With the help of
browser-based games, voluntary internet users do fun tasks
that are relative easy for humans to do but difficult to
impossible for computers, for example labeling of or locating
objects in random images found in the web. The main
challenges when creating such games are motivating users to
play the game while generating useful data, and ensuring that
the process yields unbiased results. In the recent years, several
applications of games with a purpose have been discovered
such as tagging images[5], music[9], collecting common sense
data.[8]
Several games have been designed such as Foldit, Eterna
which applications in biology problems such as protein
folding. Phrase Detectives is an "annotation game" geared
towards lovers of literature, grammar and language. It lets
users indicate relationships between words and phrases to
create a resources that is rich in linguistic information.
gwap.com[5][7] is the gaming portal which originated
games with a purpose with collection of many games
developed at Carnegie Mellon University. In this paper, we
describe another portal developed hosting games useful in
different domains. It is also useful in interlinking different
games where data collected from one game can be utilized by
other. For example, Tag graphs generated from Tag
Articles/Videos/Art game, described in the later section can be
analyzed by Label graphs game and hence produce
comprehensive analysis. Also, data collected from different
games can be utilized to draw relationships between different
objects such as for example, between articles, videos, images.
978-1-4799-0820-2/13/$31.00 ©2013 IEEE
II. GAMES
A. Similar Choices
'Similar Choices' can be used as application of games with
a purpose in E-commerce Analytics. Different products
currently active in the market are displayed to the players.
Player and his/her partner are shown a product, they have to
type in response to their product. Information about these
products is collected from online shopping websites.
Responses can be cool, ok, perfect, highly recommended, must
buy!, looks great!, not recommended etc. Each of these
responses have a particular score. Different responses can
have same scores and responses with marginal difference e.g.5 points are also accepted but player and his/her partner get
less points in this scenario as compared to when their is
perfect match. For example, must buy! can have 50 points,
cool - 40 points, perfect - 40 points, highly recommended - 40
points, ok - 25 points. If player and his/her partner's response
has a SIMILAR SCORE, their choice is considered to be identical
and they both get certain points. In this way cumulative points
of a certain product can be calculated based on user responses
and its rating/popularity as compared to other products can be
established. Also, data such as feedback, suggestions,
questions regarding products can be collected via this game.
Below figures illustrate the example of book review done
through Similar Choices.
Figure 1: Oxford Dictionary getting displayed for getting players feedbacks
Figure 2: Book reviews: Feedbacks to various books provided by the players
54
CGAMES 2013
B. Label
L
Graphs
n of games wiith a purpose in
Label graphs is application
g
analysiss is a major neeed
stattistical analysis. Statistical graphs
in almost every field from meedical, astronoomy, to softwaare
devvelopment. Toons of statisttical plots/datta needs to be
anaalyzed on a reegular basis. For
F example, companies
c
whiich
hosst dynamic weeb systems neeed to perform
m thorough looad
testting of systemss on a regular basis which reequire analysis of
respponse time ploots, in hospitalls cardio graphhic plots needs to
be analyzed etc. This
T game disp
plays plots andd player are askked
p
Some prroperties incluude
to enter the propperty of the plots.
p
cyclicc, finding verttex
incrreasing, decreasing, peak, periodic,
havving highest deegree etc. If pllayer and his/hher partner agrree
on a property, theey get points. Thus, labelingg the graphs with
w
s
analyysis.
their properties caan aid a lot in statistical
The 18th International Conference on Computer Games
C. Tag Earth
Tagg Earth is a gam
me which ask users
u
to tag sattellite images
obtaineed via Google Earth. Tags help
h
in identiffying various
landmaarks such as flyyovers, skyscraapers to name of
o some local
shop inn the region. Player and hiis/her partner are matched
accordiing to their geeographic locattion. Tags alsoo belong to a
differennt genre in whhich if player feel,
f
he and hiis/her partner
noticedd something strange
s
in thee image, theyy label it as
strangee.
Figurre 6: A Satellite im
mage getting displaayed
Figure 3: Graph depicting
d
Server reesponse time gettinng displayed
Figgure 7: Tags givenn to the satellite im
mages inclusive off strange tag
Figuure 4: Graphs labeled with propertiess such as increasinng, peak, periodic
A
Above
figures shhow, Graphs obtaiined from differennt scenarios suchh as
loadd testing, astronom
mical observations and analyzed via Label
L
Graphs.
It can also be usefful in fields such
h as Network Annalysis, Energy Grid
G
Optiimization, Fraud Detection,
D
Social Graph
G
Analysis.
An example
e
of social Network graph that can be analyzedd
D. Taggging Complex Objects – Articles/ Videos/ Art
A
User are asked to enter multiple labels, maxim
mum three for
v
or news
one viddeo or a news article. As objjects such as video
articless contains lot of
o informationn, matching sinngle common
keyworrds for these obbjects can be a bit difficult which
w
will be
a bit unnfair for the plaayers. So, users are asked to enter
e
three or
less, beest keywords for a given object.
o
For maatching, these
keyworrds might not be
b same stringgs, they are anaalyzed if they
belong to the samee context, whhether are synonyms and
matcheed accordinglyy, also a subsset of keyworrds might be
matcheed. If for a givven object moore keywords match, more
points player
p
and his partner gets. Players
P
can alsso choose the
categorry for their laabels. They caan communicate with each
other too mutually agrree on a categgory via chat interface.
i
For
examplle, if they wishh to label withh emotions theyy can choose
emotionns. Categories can be emotioons, generic.
A
i)TAG Articles
Fiigure 5: Social Nettwork Graph
Anaalyze a friendship network between members of a clubb. Properties that can
be analyzed such thhat total number of members, most
m
popular/frienndly
mem
mbers.
Some exampless of work in this area include Algoraph[3]
A
whhich
solvves problems such
s
as graph pebbling, covvering, bandwiddth
probblems. Games foor social network
ks analysis incluude [4][6].
978-1-4799-0820-2/13/$31.00 ©2013 IEEE
F
Figure
8: A Link to
t A News Article getting Displayedd to Tag
55
CGAMES 2013
The 18th International Conference on Computer Games
Figure 8, shows tagging example for news article, users read the
article at link and tag it accordingly. Tags are ranked via Offlimit
words. Old/Historical Newspapers/Articles are commonly available
as Scanned pdfs. Tags generated from game can increase their
accessibility via search. Figure 9, illustrates tagging of historic
newspaper.
Figure 13: Paintings with various labels
Figure 9: An Old/Historical newspaper article getting displayed to tag
ii)TAG Videos
Figure 10: A Linking Park song video getting displayed to the user to Tag
It can also be useful in News/Video Analytics and can be used to
obtain feedback/reactions corresponding to News Articles/Videos
such as excellent, cool, happy, sad etc. Emotion Tags/ Feedback Tags
obtained via games can help in comprehensive sentiment analysis of
various objects such as images from image sharing websites such as
flickr, news articles from different News websites, videos from video
sharing
websites,
of
social
networks,
e-communities
present. Tags obtained can be used to obtain connection between
objects such as news articles, videos.
As Games, also serve the purpose of feedback system, they can be
used to generate ranking of various objects such as products available
in the market belonging to different categories and generate
ranking/comparison of products of similar category belonging to
different brands. Ranking of news articles, videos can be generated
based on quantity of different feedbacks obtained such as excellent,
good, ok etc.
As Games use the concept of off-limit in tagging, tags obtained for
various objects such as videos, news articles etc can be ranked,
common tags which are high in number and have become off-limit
words earlier obtain high rankings as compared to other tags
generated by players.
Figure 11: Different Tags Identifying context such as pop, music are identical,
snake, python are identical. Also tags can be both generic labels as well as
those depicting Reaction/Emotions
iii)TAG ART
Tag art is a game which ask users to assign tags to pieces of
art such as paintings. Tags can be a generic label or a review
label such as excellent, okay, etc. As paintings are complex,
users can enter multiple keywords as in case of videos/articles.
Figure 14: Tag Graph
Tags which are higher ranked as obtained from games form the
vertex of the graph and which occur in more news articles, videos,
images form a network. Tag Graph is shown in figure 14. In this way,
deeper relationships between news articles, videos, images can be
obtained based on complexity of the graph. Portal also features a
game for assigning emotion tags to images obtained from image
sharing sites such as flickr.
E. Pick Quality
Figure 12: A painting getting displayed to tag
978-1-4799-0820-2/13/$31.00 ©2013 IEEE
Pick Quality is a fun game utilizing ESP Game framework. In
Pick Quality, images of superheroes from various movies, games,
comics are shown to the player, they have to give the quality of the
superhero, if both players give the same quality, they get points. After
sometime, common/predominant quality become off-limits and
players will have to type in a different quality of the superhero which
56
CGAMES 2013
addds challenge to thhe game. Figuree 15 shows tags generated via Pick
Quaality Game.
a)
The 18th International Conference on Computer Games
IIII.
PORTAL DESIGN
Porrtal also has othher useful featuures described below which
gives complete enteertainment/rew
warding experiience. Game
playerss can also obtaain credits/poinnts which can be
b utilized in
below features suchh as online shopping,
s
thuus acts as a
motivattion for games with a purpose.
A.Classsic Games
b)
F
Figure
18: Classic Games
G
getting Dissplayed to Play
Figure 15: Tagging Examplles from ‘Pick Quaality’
L
Program
mming Game
F. Learn
Learn Progrramming is a game that stuudents can use to
learrn programminng and preparre for variouss exams such as
GR
RE, job intervviews. It con
nsists of quesstions regardiing
geoometry questioons, programming questions to predict outpput
as shown in Figgure 16, spot errors in thee program, woord
puzzzles to strengtthen vocabulary
y, common maath puzzles.
As players, pllay this game in their free tiime, they will be
able to revise im
mportant conceepts in fun maanner. These are
a
p
and
d often tends to
t get left out as
inteegral part of preparation
studdents find it monotonous
m
an
nd hence, donn't give time. As
theyy play with a partner, comp
petitive spirit gives
g
enthusiassm.
Theey can also connect, chat, discuss, shaare via chattiing
inteerface.
Classsic games suchh as Super Marrio, Sonic, Tetrris etc
can be played
p
at the portal,
p
adding too the fun optionns.
B.ONLIN
INE SHOPPING
i)Gaminng Store
Figure 19: Soome products at gaaming store
Portal features a gam
ming store whhere one can shop
s
for cool
ms
gamingg products/item
F
store
ii)Eco- Friendly
Figure 16: Javva Question to Pred
dict Output displayyed via Game
Figure 20: Soome products at Ecco-Friendly store
Fiigure 17: Player annd his partner can see
s their answers and
a problems in thhe
en
nd
978-1-4799-0820-2/13/$31.00 ©2013 IEEE
Porrtal also containns an eco–frienndly store wheere cool, ecofriendlyy objects can be purchased. One useful feeature of this
store iss that funds obbtained from this
t
eco- frienndly store are
used too support eco-- friendly projects hosted at eco-friendly
organissations such as Carbonffund.org, Plaant a Tree
organizzations.
57
CGAMES 2013
The 18th International Conference on Computer Games
V. SURVE
EY
C.CHATBOX
motions, Pick
A suurvey conducteed reported thaat Identical Em
Quality,, Tag Art were more
m
popular as compared to Sim
milar choices .
Also Ecco-Friendly shopp provided a goood incentive foor players and
contribuuted to a good caause.
VI. SYSTEM
S
IMP
PLEMENTATIION
Currenttly Gaming syystems have been
b
implemennted in PHP
taking scaling issuess under considderation. Java version also
hasbeenn developed.
Figure 21: ChattBox where playerrs can communicatte
Portal containss a chatbox wh
here players acctive at the porrtal
cann communicatee share with eaach other, givee feedback aboout
the portal. They can thus, also
o connect to each
e
other whhile
playying.
VII.. OTHER APPLICATION
A
NS OF GAME
ES WITH A
PURPO
OSE
Above Games
G
can be utilizzed to solve probleems such as –
1)Name
ed
Entity Recognition
Identifyy people, compannies, organizatioons, cities, geogrraphic features,
and otheer typed entities within webpagees and text docum
ments/content.
D.L
Latest News/Off
ffers
out gaming woorld, technologgy,
One can get latest news abo
wd sourcing prooject informatiion
eveents such as exhibitions, crow
andd other offerrs/promotions related to games,
g
produucts
avaailable at onlinee shops via this portal.
word/Term
2)Keyw
IV. DISC
CUSSION
The concept of
o using digital games as inccentive for hum
man
mputation seemss very beneficiall. Observing som
me statistics aboout
com
gam
mes on the sociaal network platfform Facebook supports
s
the aboove
stattement. According to [21] “B
By September 2010 players on
Faccebook played more than 900
0 million hourss per month [119].
Connsidering the labbelling rate reported by von Ahnn [5] this would be
~199 million labels per hour. Hum
man computation games are a veery
speccial type of gam
mes, they have repetitive
r
game mechanics. Gam
mes
suchh as FarmVille also have a rep
petitive game play and were veery
succcessful.” Humann computation games
g
are designned around the taask
theyy try to solve. The main idea thereby is to make
m
a task moore
apppealing by addingg game elements. Approaches trrying to make suuch
monnotonous thingss more interestting/enjoyable by
b applying gaame
mecchanics are term
med as gamificaation. Use of stoorylines will maake
hum
man computationn games more interesting. Alsso, other conceepts
suchh as providing badges to playeers can be usefu
ful. Another useeful
featture can be conttent unlocking in
i case of objeccts such as videeos,
painntings, articles such as showin
ng historical impportant newspapper
articcles, if players has progressed
d to a certain leevel and earnedd a
baddge , after taggging certain nu
umber of objects. Also, addiing
inceentives such as features
f
of gamiing portal wheree games are hosted
desccribed above, such
s
as online shops, classic games, will maake
hum
man computationn games more in
nteresting.
Porrtal is available at
a - sites.google
e.com/site/kkooolgames/
978-1-4799-0820-2/13/$31.00 ©2013 IEEE
Extrraction
Extractt important terrms and topic keywords froom webpages
and texxt documents/content.
fication
3)Topicc Categorisatioon/Text Classifi
Automaatically topic-classsify web pages and text documeents/content
58
CGAMES 2013
4)A
Author Extractiion
Auttomatically extraact author inform
mation from new
ws articles or bllog
postts.
The 18th International Conference on Computer Games
G
Adam Perer, Tal Daniel, Ohadd Greenshpan, Itai Turbahn: Guess
[4] Ido Guy,
Who? Ennriching the Sociial Graph throughh a Crowdsourcinng Game. ACM
Conferennce on Human Factors in Com
mputing Systemss (CHI 2011).
Vancouvver, Canada. (2011).
[5]]Luis von Ahn, and Lauura Dabbish. ESP: Labeling Images with
w a Computer
Game. AAAI
A
Spring Sym
mposium: Knowlledge Collection from Volunteer
Contribuutors, page 91-98. AAAI,
A
(2005
[6]Moham
med Shehab, Mooo Nam Ko, and Hakim Touati, Social networks
profile mapping
m
using gam
mes, WebApps'12 Proceedings of thhe 3rd USENIX
conferencce on Web Applicaation Developmennt
[7] http:///www.gwap.com/ggwap/
Automatic Langguage Identificcation
5)A
Dettermine the lannguage that any
y text or web-bbased content was
w
writtten in.
[8]Luis von
v Ahn, M. Keddia, and M. Blum. Verbosity: a gam
me for collecting
common--sense facts.CHI '06: Proceedings of the SIGCHII conference on
Human Factors
F
in computting systems, pagge 75--78. New York,
Y
NY, USA,
ACM Preess, (2006)
[9]Edith L. M. Law, Luis von
v Ahn, Roger B. Dannenberg, andd Mike Crawford
TagATunne: A Game for Music
M
and Sound Annotation.
A
ISMIR
R, page 361-364.
Austrian Computer Societyy, (2007)
[10] Robbert Speer, Catherrine Havasi, and Harshit Surana .U
Using Verbosity:
Commonn Sense Data from Games with a Purrpose. FLAIRS Coonference, AAAI
Press, (20010)
[11]Clauudiu Cristian Muusat, Alireza Ghasemi, Boi Faltiings: Sentiment
Analysis Using a Novel Human
H
Computattion Game. ACL 2012 , People's
Web Meeets NLP Workshopp, 2012.
Sentiment Anallysis
6)S
on extraction, Identify facts and
a
Othher problems can be relatio
Subbject-Action-Object relations within webppages and text
t
doccuments/content. For example – subject –china, action- willingg to
buyy, object – bondds from debt criisis nations, otheer problem can be
conntent/data extraction such as prroduct prices, product
p
description
from
m the webpage. Above
A
problemss have been derivved from [24].
mes is, that theyy can be used to
Useful featuure of the gam
solvve above probllems for other objects as welll as such as appart
from
m text webpagges, articles, other
o
complex objects such as
viddeos, sound recordings, imagees, art and thuss, be able to draaw
com
mprehensive coonnections betw
ween them.
VIII. REFERENCES
[1] Segaran,
S
T. Prograamming Collectivee Intelligence: Buiilding Smart Web 2.0
Appplications. O'Reillyy Media, Inc., Beijjing, 2007.
S
J. Thee Wisdom of Crow
wds: Why the Manny Are Smarter Thhan
[2] Surowiecki,
the Few
F and How Collective Wisdom Shapes
S
Business, Economies,
E
Societies
and Nations, London, 2004
[3] http://graph.compu
h
utinggames.org/
978-1-4799-0820-2/13/$31.00 ©2013 IEEE
[12]Lisa Pearl, Mark Steyvvers, Identifying emotions,
e
intentioons, and attitudes
in text using
u
a game witth a purpose. CAAGET '10, Prooceedings of the
NAACL HLT 2010 Worksshop on Computattional Approaches to Analysis and
T
Generatioon of Emotion in Text
[13]http://www.thepcf.org..uk/your_paintingss
wikischolars.colum
mbia.edu/Network+
+Graph+Navigat
[14]http://mediamapping.w
ion
[15]http://newspaperarchivve.com/
[16] S. Matyas,
M
C. Matyaas, and C. Schliedder, “Designing a Location-based
Game foor the Collection of
o Geospatial Dataa: The CityExplorrer Case Study,”
ACE ’08: Proceedings of
o the Internationnal Conference on
o Advances in
T
Yokohhama: ACM Presss, 2008, pp. 244Computeer Entertainment Technology,
247
M
Playful geoospatial data acquissition by location-bbased gaming
[17] S. Matyas.
communiities. The Internatiional Journal of Viirtual Reality, 6(3)::1–10,2007.
[18] http://code.google.com
m/p/esp-game-courrse-project/
Facebook, “Faceboook Games Statisticcs,” 2010, p.
[19] AllF
http://ww
ww.allfacebook.com
m/facebook-gamessstatistics-2010-099.
[20] Mann-Ching Yuen, Linng-Jyh Chen, and Irwin King. A Suurvey of Human
Computaation Systems. IEE
EE Symposium onn Social Computiing Applications
(SCA'09,, in conjunction wiith IEEE SocialCom'09), Vancouver,, Canada, 2009
[21] Maarkus Krause and Jan Smeddnick. Human Computtation Games: a
Survey. In
I EUSIPCO '11 Proceedings
P
of thee 19th European Signal Processing
Conferennce , Barcelona, Sppain, 2011
[22]A. Simon.
S
Social herroes: games as apis
a
for social innteraction. In S.
Tsekeridoou, A. D. Cheok, K. Giannakis, andd J. Karigiannis, editors,
e
DIMEA,
volume 349
3 of ACM Internnational Conferencce Proceeding Seriies, pages 40–45.
ACM, 20008
[23] Yotaam Gingold, Etiennne Vouga, Eitan Grinspun,
G
Haym Hirsh,
H
Diamonds
From thhe Rough: Impproving Drawingg, Painting, andd Singing via
Crowdsoourcing. Human Coomputation Workshhop (HCOMP), 20012.
[24] httpp://www.alchemyaapi.com/
[25] C. Dugan,
D
M. J. Mulller, D. R. Millen, W.
W Geyer, B. Brow
wnholtz, and M.
Moore. The
T dogear game: a social bookmarkk recommender sysstem. In T. Gross
and K. Innkpen, editors, GROUP, pages 387–3390. ACM, 2007.
59