5. Games 4/13/11 www.insemtives.eu 5 > 300million downloads1 They are fun… 200 million people play casual games2 >15 million daily active players3 Casual Games Market Report, 2007 http://www.casualgamesassociation.org/news.php Facebook application statistics, Feb 2011 5
6. Casual games Steep learning curve Fast game play: little time effort required Simple implementation (simple interface and graphics) Low hardware efforts (usually browser or mobile app) Low bandwidth requirements Mass audience www.insemtives.eu 6 4/13/11
10. OntoGame Players paired randomly and anonymously Best strategy to get points: truthful answers 2 player live mode or Community matching mode Skip Limited amount of time Cheating: Anonymity Pre-recorded challenges Generic gaming platform Derive formal representations of the data All data exported as linked data www.insemtives.eu 10 4/13/11
20. SeaFish – Collected Data 3 Different data sets An online survey Feedback from Siegen, Trento 14456 Modeling decisions 931 Game rounds 548 Generated annotations 4/13/11 www.insemtives.eu 19
26. Lessonslearnt Task selection Not too easy nor too difficult, suitable for a broad audience, not too many game rounds (3-5) Simple challenges (nottoomanystages, etc.) Limitedspace of choice Allow fast gameplay Game fun: user interface and social factor Compromise: usability and appealing design It isnot trivial to make an interesting user interface Competition and reputation Ranking is a good motivator Sociability Playing against other people ismotivating Partner! Knowledge corpora Interesting domain, structure and size of the corpora www.insemtives.eu 25 4/13/11
28. OntoGame API API that provides several methods that are shared by the OntoGame games, such as: Different agreement types, e.g. Selection Agreement Input matching (e.g. majority) Game modes (double player, single player) Player reliability evaluation. Player matching, e.g. finding the optimal partner to play. Resource( = data needed for games) management. Semantic data extraction of stored game data. http://insemtives.svn.sourceforge.net/viewvc/insemtives/generic-gaming-toolkit www.insemtives.eu 27 4/13/11
33. References Luis von Ahn. Games With A Purpose. IEEE Computer Magazine, June 2006. pp 96-98. Luis von Ahn and Laura Dabbish. Designing games with a purpose. Communications of the ACM, 2008. Katharina Siorpaes and Elena Simperl: Human Intelligence in the Process of Semantic Content Creation, World Wide Web Journal (WWW), Volume 13, Issue 1-2, March 2010. Katharina Siorpaes and Martin Hepp: Games with a Purpose for the Semantic Web. IEEE Intelligent Systems, Vol. 23, No. 3, pp. 50-60, May/June 2008. CasualGame Association White Paper: http://www.casualgamesassociation.org List of games: http://www.insemtives.eu/games.php www.insemtives.eu 32 4/13/11
35. Tasks and deliverables All tools are available on http://sourceforge.net/projects/insemtives/ Information about tools is provided on http://insemtives.eu/community/tools 4/13/11 www.insemtives.eu 34
36. Work plan view Months 24 12 18 30 36 6 0 Tasks SEMANTIC GAMES D4.1.1 Requirements and Design of a Generic Gaming Toolkit and API D4.1.2 Generic Gaming Toolkit and API implementation D4.1.3 Games Task 4.1 Genericgaming toolkit UIBK WEB SERVICE ANNOTATION CHALLENGE D4.2.2 Human‐driven annotation tools for Web services D4.2.1 Human‐driven annotation tools for Web services Task 4.2 Human‐driven Annotation Tools SEEKDA/ONTO D4.2.4 Human‐driven media annotation tools BOOTSTRAPPING TOOL AND L!NKS SEMI AUTOMATIC IMAGE ANNOTATION D4.2.3 Human‐driven media annotation tools D4.3.2 Bootstrapping tools for image files Task 4.3 Bootstrapping Tools UNITN D4.3.1 Bootstrapping tools for image files L!NKS MANUAL IMAGE ANNOTATION AND SEARCH D4.4.2 Search and Navigation tools Task 4.4 Search and Navigation Tools ONTO D4.4.1 Search and Navigation tools
37. Guidelines Timed response Score keeping Player skill level High score lists Randomness Random player pairing Player testing Repetition Taboo outputs www.insemtives.eu 36 Luis von Ahn. Games With A Purpose. IEEE Computer Magazine, June 2006. pp 96-98. 4/13/11
38. How to design your own game Specify output Identify input Choose type of game and define game play Based on previous decisions, define game play and adapt underlying game ontology Adapt or define export algorithm Evaluate output www.insemtives.eu 37 4/13/11
39. OntoTube: designflaws Consensus findinghard: toomanyoptions to choosefrom. Gameplaycomplex: toomany different stages (8 questions). Toocomplexinterfaces: playershave to choosefromwidelyvaryingUis in different stages of the samegame. Videos toolong: for a fast gameplay, usersshouldnotberequired to watch the completevideo. Socialfactor: the interactionwith the partneristoolow – players do notenjoy the gamebecause the emphasis on the socialcomponentismissing. www.insemtives.eu 38 4/13/11
40. Challenges Identifying suitable tasks in semantic content creation Designing games Designing a usable, attractive interface Identifying suitable knowledge corpora Preventing cheating Defusing typical pitfalls of conceptual modeling Distribution of labor Fostering user participation Deriving formal representations Scalability and performance www.insemtives.eu 39 4/13/11
Explain semantic web ( make the web more intelligent, providing homogenously structured meta data, allow reasoning) Two parted problem: Semantic web needs dataSemantic Web needs Machine processable Meta data; Backbone of the Semantic WebNot a one time effort, Continuos process, Web highly dynamic entityTasks that require human input - Difficult for machines, easy for human “computation”Captchas: distorted images to distinguish between maschinesHigh level Semantic Annotation: Audio / Video / Images / ConceptualizationOntology aligment / evaluation / learning / Building hierrachies / reuse, propertiesSo, If we need human => howto motivate them
people play games, a lot, growing sector, getting increasingly popularcollected some numbers, speak for themselvesEspecially intersting for our case: casusal games (no steap learning curve, little hardware requirements, high availability via web) and quiz gamesMass audience:High potential => free labour, news.php provide input, unused cycles
Second part of Talk, introduce games*advertisment
Small scale evaluation, Same level of consensual answers, 80% of the resulting alignments were correctAt time of evaluation: 190gamerounds, 32 concepts, 16 players, average answers per player 23,56, number of records recorded 882Difficulty: 31.2% considered it difficult, majority would not want to play it againFun: huge majority didn’t consider it fun, majority don’t want to play againLearned => next version, compiled all we learned => SLODTheLink