Presentation on theme: "In765 Knowledge Networks: A Structural Study of Networks Judith Molka-Danielsen Molde University College"— Presentation transcript:
In765 Knowledge Networks: A Structural Study of Networks Judith Molka-Danielsen Molde University College email@example.com http://home.himolde.no/~molka 2005
Types of Networks.. Social Networks Friends, families, colleagues Logical Resources web-pages,P2P-Gnutella Telephone Wireline, mobile Transport Roads, railroads, airline, electricity Economic Firms, markets, organization Physical Resources Internet IP routers
Why Study Networks? Research Areas Availability and vulnerability of services: electric, telephone, air connections, etc. Preventing or stopping of viruses on data networks. The importance of weak ties: connectivness to the core, finding a job, finding a web page. The characterization of network structure and the role of hubs in the spreading an idea, or proliferation of a product, and managing organizations.
Former Research Random Network Theory –Erdös & Rényi (1960) Six Degrees of Separation –S.Milgram (1967) Cluster Coefficient –Small Worlds – Watts & Strogatz (1998) Hubs and Scale Free Networks – Albert, Jeong, & Barab á si (1999) Hubs in Social Networks – Malcolm Gladwell (2000)
Random Networks Erdös-Rényi model (1960) - Democratic - Random Pál Erdös (1913-1996) Connect with probability p p=1/6 N=10 k ~ 1.5 Poisson distribution
Six Degrees of Separation Nodes: individuals Links: social relationship (family/work/friendship/etc.) S. Milgram (1967) Social networks: Many individuals with diverse social interactions between them. John Guare (1980) Six Degrees of Separation
Cluster Coefficient Clustering: My friends will likely know each other! Probability to be connected C » p C = # of links between 1,2,…n neighbors n(n-1)/2 C friends = 15/ [6(5)/2] = 100%
Hubs in Networks 200 million searches each day More than 2300 searches per second In 88 languages 3.2 billion web pages indexed. 10 000 super computers perform the searches.
Do we find Hubs in Social Networks? Yes. Most influencial Access to the most information Impacts others decisions most Have the most power
Who do you know? (similar to a study by Malcolm Gladwell, 2000) Bjørnstjerne Bjørnsons Vei Alme Jørund Andenes Aud Andestad Reidar Bakke Gerd Inger Bergseth Egil Bergtun Lill Eldrid Bjøringsøy Karl Magnar Bjørkly Jorunn Bjørkly Åsa Bjordal Bjørnebo Solveig Randi Midtbø Broks Vivi-Annie Brokstad Jon Drageseth Dagfinn Dyrli Janne Merete Døving Ellen Eilertsen Gudny Flø Jorunn Marie Fylling Lars Kristen Tovan Gjære Arne Gjære Guro Wiersholm Gjære Vibeke Wiersholm Grønbugt Rutt Grønset Erling Rune Gudbrandsen Åste Einbu Gøncz Geir Janos Göncz Arne Hansen Helge Hansen Sissel Helde Marit Illøkken Henriksen Line Hjelmsøt Maria Hoem Jermund Hofset Siv Jenset Grete Jenset Torbjørn Jordet Birgit Kanestrøm Andreas Julshamn …
12- 15 1 8-113 4-75 0-313 151101 91 81 62 52 31 24 16 03 A = number of persons known on the list. B = number of persons (nodes) that person A knows. AB AB Who do you know?: survey to faculty gruppert 22
15- 23 1 7-141 5-62 2-48 0-136 231141 61 51 44 31 23 116 020 AB AB Who do you know?: survey to students gruppert 48 A = number of persons known on the list. B = number of persons (nodes) that person A knows.
Scale Free Networks and Power Laws by Albert, Jeong, Barabasi.
Collaboration Among Researchers Networks have diverse nodes and links are -computers -routers -satellites - researchers -phone lines -TV cables -EM waves - co-authorship
Unique co-author link distribution – researchers represented individually
Informatics Institute Cluster: researchers and co-author links
Health Institute Cluster: researchers and co-author links
Social Sciences Institute: researchers and co-author links
Economics/Logistics Cluster: researchers and co-author links
Economics Institute Cluster: researchers and co-author links
Connected Network Tree of researchers and co-author links
Conclusions Network of researchers at HSM is a Scale Free network. (existance of hubs, clustering coeffiencient) Co-authors are not chosen randomly. Co-authorship & Publication count: (cannot claim causality) –Average # of co-author per paper is the same regardless of the total # of publications per author. (does not help) –Average # of unique associations is related to a total # of publications per author. (helps) Role of “ connectors” (nodes with a high # of external links) are important –They often have high publication counts. –They have more external contacts. –They are more likely to hold a joint appointment (again not causal).