Ce diaporama a bien été signalé.
Nous utilisons votre profil LinkedIn et vos données d’activité pour vous proposer des publicités personnalisées et pertinentes. Vous pouvez changer vos préférences de publicités à tout moment.
Tagging structure in a protein-protein interaction network, a co-authorship network and the (English) Wikipedia Proteins M...
G. Palla, I. J. Farkas, P. Pollner, I. Derényi, T. Vicsek  Fundamental statistical features and self-similar properties of...
There is no clearly separated group of “most popular tags” Transition between popular and less popular tags is  continuous...
There is no clear group of “most heavily tagged nodes” Transition between strongly and less strongly tagged nodes is  cont...
On the large scale there is no “link saturation” within a topic In fact, link density within a large topic is  almost the ...
CFinder.org A B C D E F G H J Meaningful removal of loops from the category hierarchy How to achieve tree structure (DAG) ...
CFinder.org Meaningful removal of loops from the category hierarchy How to achieve tree structure (DAG) with the lowest nu...
Gergely Palla  Illés Farka s  P é ter Pollner  Imre Derényi   Tamás Vicsek
http://CFinder.org -- with network data and free analysis software We thank for support from:
Prochain SlideShare
Chargement dans…5
×

Structure of tagging on Wikipedia pages and other networks -- Wikimania 2010 -- Gdansk

  • Identifiez-vous pour voir les commentaires

  • Soyez le premier à aimer ceci

Structure of tagging on Wikipedia pages and other networks -- Wikimania 2010 -- Gdansk

  1. 1. Tagging structure in a protein-protein interaction network, a co-authorship network and the (English) Wikipedia Proteins Math co-authors Wikipedia articles Network Nodes + Links Node tags Directed Acyclic Graph Gergely Palla , Illés J. Farka s , P é ter Pollner , Imre Derényi, Tamás Vicsek Category tree FIFA World Cup FIFA World Cup Players Classification of co-authored papers Combinatorics Graph theory Biochemical functions Growth Cell growth E ötvös University and Hungarian Academy of Sciences (Budapest, Hungary) Interactions (“MIPS”) Co-authorships (“MathSciNet”) Hyperlinks in Wikipedia CFinder.org
  2. 2. G. Palla, I. J. Farkas, P. Pollner, I. Derényi, T. Vicsek  Fundamental statistical features and self-similar properties of tagged networks   New Journal of Physics   10 , 123026 (2008)  http://CFinder.org --> Publications Full version:
  3. 3. There is no clearly separated group of “most popular tags” Transition between popular and less popular tags is continuous Portion ( x ) of all nodes in the network Probability that a given tag and its descendants label a portion x of all nodes CFinder.org
  4. 4. There is no clear group of “most heavily tagged nodes” Transition between strongly and less strongly tagged nodes is continuous Number of tags ( n ) on a node (protein, author, wiki article) of the network Probability that a node has n tags CFinder.org
  5. 5. On the large scale there is no “link saturation” within a topic In fact, link density within a large topic is almost the same as outside CFinder.org Number of links among these nodes Number of Wikipedia articles selected from those labeled with “Japan” and descendant terms Maximum possible Wikipedia average
  6. 6. CFinder.org A B C D E F G H J Meaningful removal of loops from the category hierarchy How to achieve tree structure (DAG) with the lowest number of removals Example (Oct.2007): <ul><li>Finding all loops is easy. </li></ul><ul><li>But which of their links should be removed? </li></ul><ul><li>Goals: </li></ul><ul><li>Remove lowest possible number of links </li></ul><ul><li>- Smallest “damage” to existing category hierarchy </li></ul>subcategory subcategory Category:Urdu Category:Hindustani
  7. 7. CFinder.org Meaningful removal of loops from the category hierarchy How to achieve tree structure (DAG) with the lowest number of removals Example (Oct.2007): <ul><li>Identify “loop subgraph” </li></ul><ul><li>by iteratively removing nodes with 1 link </li></ul><ul><li>(2) Iteratively remove “least important” links </li></ul><ul><li>from the loop subgraph </li></ul><ul><li>(3) Add non-loop links again </li></ul>A B C D E F G H J subcategory subcategory <ul><li>Finding all loops is easy. </li></ul><ul><li>But which of their links should be removed? </li></ul><ul><li>Goals: </li></ul><ul><li>Remove lowest possible number of links </li></ul><ul><li>- Smallest “damage” to existing category hierarchy </li></ul>Category:Urdu Category:Hindustani
  8. 8. Gergely Palla Illés Farka s P é ter Pollner Imre Derényi Tamás Vicsek
  9. 9. http://CFinder.org -- with network data and free analysis software We thank for support from:

×