video game industry modeled by complex networks
DESCRIPTION
Presented By Tony Morelli. Video Game Industry modeled by complex networks. Outline. Intro/Problem description Visual Network Representations Numerical Network Representations Questions/Comments. INTRO. Has the organization of the video game industry changed in the last 20 years? - PowerPoint PPT PresentationTRANSCRIPT
VIDEO GAME INDUSTRY
MODELED BY COMPLEX
NETWORKS
Presented By Tony Morelli
Outline Intro/Problem description Visual Network Representations Numerical Network Representations Questions/Comments
INTRO Has the organization of the video game
industry changed in the last 20 years?ConsolesGame TitlesProducersDevelopers
Consoles to Analyze Group A – Classic Consoles
Atari 2600 (1977)
Consoles to Analyze Group A – Classic Consoles
Atari 2600 (1977)Nintendo Entertainment System (1983)
Consoles to Analyze Group A – Classic Consoles
Atari 2600 (1977)Nintendo Entertainment System (1983)Sega Master System (1985)
Consoles to Analyze Group B – Current Consoles
XBOX 360 (2005)
Consoles to Analyze Group B – Current Consoles
XBOX 360 (2005)Playstation 3 (2006)
Consoles to Analyze Group B – Current Consoles
XBOX 360 (2005)Playstation 3 (2006)Nintendo Wii (2006)
Where is the data? www.games-db.com (Classic Consoles) www.ign.com (Current Consoles) Console,Title,Developer,Publisher
Background/Related Work Comparison has not previously been
done Need to investigate techniques for
network comparison
Methods of Comparison Graphical Numeric
Soft Drink Industry
Soft Drink Industry
Seed Industry Consolidation
Seed Industry Consolidation
http://www.youtube.com/watch?v=nBBXLZWyXBQ
Non-Scientific Network
Graphical Representation Use Colors and Sizes Use Pajek to Generate Use morphing animation to show
changes from classic vs current
Graphical Difference How do the two graphs differ visually?
Hypothesis – Current consoles have less producers with more content than classic.
Numerical Analysis Several Studies have been done
showing numerical analysis of networks Important to find metrics and
comparison methods
Network Topologies, Power Laws, and Hierarchy
Published June 2001 Analyzes Topology Generators
Network Topologies, Power Laws, and Hierarchy
Internet researches had usedGT-ITMTiers
Generated a simulated internet to test and analyze
Network Topologies, Power Laws, and Hierarchy
Faloutsos found:Internet’s degree distribution is power lawGenerated topologies are notTherefore generated topologies are a poor
choice to run studies on
Network Topologies, Power Laws, and Hierarchy
This paper focusses on a comparison ofDegree-based generators
○ Degree Distribution is the focusStructural generators
○ A hierarchical structure is the focus
Network Topologies, Power Laws, and Hierarchy
Found Degree Based Generators are betterBased on the metrics they usedWhat are these metrics?
Network Topologies, Power Laws, and Hierarchy
MetricsExpansion
○ “The average fraction of nodes in the graph that fall within a ball of radius r, centered at a node in the topology
Network Topologies, Power Laws, and Hierarchy
MetricsResilience
○ How tolerant is the network to failures?○ Cut a single link in a tree
No longer connected○ Cut a single link in a random graph
Probably OK○ Average cut-set size within an N node ball
around any node in the topology
Network Topologies, Power Laws, and Hierarchy
MetricsDistortion
○ Take a random node and all nodes connected to it within n hops
○ Create a spanning tree on this subgraph○ The average distance between vertices that
are connected in the original subgraph is the distortion
Network Topologies, Power Laws, and Hierarchy What metrics will I use from this paper?
Expansion seems goodDistortion and Resilience probably will not
be used.
Comparison of Translations How accurate are software based
translators? Portuguese->Spanish Portuguese->English
Comparison of Translations Translators compared
Human translatedFree TranslationIntertran
Comparison of Translations Methods
Model translated text as a directed graphNodes connected together based on
sequence of appearance in translationThe 2 machine translated networks
compared to the human translated network
Comparison of Translations Metrics
In-degree○ Frequency a word was the second word
Out-degree○ Frequency a word was the first word
Clustering Coefficient○ How much does the graph cluster together
Comparison of Translations Results
Closer the In-Degree - More accurate translation
Closer the Out-Degree - More accurate translation
Comparison of Translations
OD ID CC OD ID CCHuman 0.93 0.94 0.44 0.99 1.0 0.44
Apertium 0.97 0.98 0.93 1.01 1.03 0.85
Intertran 0.76 0.80 0.58 0.98 1.48 1.01
ResultsAvg Pearson Coefficient Avg Angular Coefficient
Comparison of Translations Results
Comparison of Translations Which metrics to use?
In-degree – Not relevantOut-degree – Could be usefulClustering Coefficient - Useful
Food-web structure and network theory
Are food web networks small world or scale free?
Food WebsRelationships in ecosystems
○ Who eats who16 food webs26-172 nodes in each web
Food-web structure and network theory
MetricsAverage shortest path length between all pairs
of speciesClustering CoefficientAverage fraction of pairs of species one link
away from a species that are also linked to each other
Cumulative degree distributionConnectance
○ The fraction of all possible links that are realized in a network
Food-web structure and network theory
ResultsSome characteristics met the standards for
small world and scale freeClustering was low
○ Could be because of network size
Finding the Most Prominent Group in Complex Networks Group Betweenness Centrality
Used to evaluate the prominence of a group of vertices
Might be time consuming to evaluate
Finding the Most Prominent Group in Complex Networks The study evaluates quick methods of
finding the most prominent group
Finding the Most Prominent Group in Complex Networks 2 algorithms
Heuristic SearchGreedy Choice
Finding the Most Prominent Group in Complex Networks 2 algorithms (Lots of math)
Heuristic Search○ Fastest
Greedy Choice○ Most accurate
Finding the Most Prominent Group in Complex Networks Useful to this project?
Video game network is probably too small to benefit from either method
Statistical Methods of Complex Networks
Average Path Length Clustering Coefficient Degree Distribution Spectral Properties
Directly related to the graph’s topological features
Statistical Methods of Complex Networks
Metrics UsedAverage Path Length
○ Not very useful for Video Game networkClustering Coefficient
○ Will useDegree Distribution
○ Will useSpectral Properties
○ Topology is already known – not useful
Apply to video games Graphical
Size and color○ Larger node has more titles tied to it○ Colorize publishers to easily distinguish○ Create an animation of classic to current
Apply to video games Numerical
Clustering CoefficientAverage out degree Expansion at each level
All will be normalized by number of titles
Work so far… Scraper has been written Written in C# Crawled the websites to gather console,
publisher, developer, title for all six consoles
Questions/Comments?
References [1] H. Kopka and P. W. Daly, A Guide to LATEX, 3rd ed. Harlow, England: Addison-Wesley, 1999. [2] J. Tidwell, Designing Interfaces: Patterns for Effective Interaction Design,. O’Reilly Media: Sebastopol, CA, USA, 2005. [3] Philip H. Howard, Visualizing Consolidation in the Global Seed Industry: 19962008,. Sustainability 2009. [4] Philip H. Howard, The illusion of diversity: visualizing ownership in the soft drink industry,. https://www.msu.edu/ howardp/softdrinks.html 2008. [5] Steven H. Strogatz, Exploring complex networks,. Nature. March 2001. [6] Zimele, Developing Community Self Reliance,. http://www.zimelecommunity.org/programs/microbanks/ 2011. [7] Hongsuda Tangmunarunkit and Ramesh Gordon and Sugih Jamin and Scott Shenker and Walter Willinger, Network Topologies, Power Laws, and Hierarchy,. ACM SIGCOMM January 2002. [8] DIEGO R. AMANCIO and LUCAS ANTIQUEIRA and THIAGO A. S. PARDO and LUCIANO da F. COSTA and OSVALDO N. OLIVEIRA, Jr. and MARIA G. V. NUNES, COMPLEX NETWORKS ANALYSIS OF MANUAL AND MACHINE TRANSLATION,. International Journal of Modern Physics Vol. 19, No. 4 (2008) 583-59. [9] Jennifer A. Dunne and Richard J. Williams and Neo D. Martinez, Food-web structure and network theory: The role of connectance and size,. Procedings of the National Academy of Sciences of the United States of America. [10] Paul Noglows, Moving Online Will Help Video Games Capture More Ad Revenue,. http://www.businessinsider.com/ 2011.