Skip to content

Latest commit

 

History

History
61 lines (51 loc) · 3.12 KB

README.md

File metadata and controls

61 lines (51 loc) · 3.12 KB

Influential-node-detection

This project is about finding the importance of the node in a community, which is formed based on the source nodes. The communities are formed using DFS traversal of the seed node. The dataset used is hep-th dataset. The following metrics are implemented for influential node detection

Libraries and Tools used

Graphia software for graph visulaization Networkx. Refer this for further information: https://networkx.org/documentation/stable/reference/introduction.html

Resources used for Reference

Centiserver is a website which contains the information about all the centrality measures used in various problem domains. Refer this for more information: https://www.centiserver.org/

Why do we need many centrality measures?

  1. optimal for one application while performs worse for a different application.
  2. features which identify the most important vertices in a given network do not necessarily generalize to the remaining vertices.

The measures that have been implemented are as follows. Description about the measures could be found in the document.

  1. Degree centrality ✅
  2. Closeness centrality ✅
  3. Betweeness centrality ✅
  4. Harmonic centrality
  5. Eigenvector centrality ✅
  6. Katz centrlaity ✅
  7. Percolation centrality
  8. PageRank ✅
  9. Freeman centrality
  10. Cross clique
  11. Dissimilarity measures
  12. HITZ algorithm ✅
  13. Degree prestige ✅
  14. Proximity prestige ✅
  15. Rank Prestige ✅
  16. Shapley–Shubik power index
  17. Voterank ✅
  18. Leaderrank ✅
  19. TARank ✅
  20. Cocitations ✅
  21. Hybrid degree centality ✅
  22. Expected force of infection ✅
  23. Reciprocal of Eccentricty ✅

These measures are adopted from this repository, where the measures are written in matlab. Here, it is implemented in python. https://github.com/xsxjtu/Complex-Network-Centrality

  1. SVDB
  2. Semilocal centrality
  3. Non backtracking

The results files attached are as follows

  1. Communities.csv - information about the communities. (no of nodes, no of sink nodes, max indegree, nodes with max indegree)
  2. HITZ.csv - HITZ algorithm (hub node nd authority node with max hub and authority score, for each community)
  3. bc_cc_kc.csv - Centrality measires(node with max betweeness centrality, closeness centrality and katz centrality with their scores for 13 communities)
  4. citation.csv - citation information (number of citations and number of papers it cite, for each node)
  5. cocitation.csv - cocited papers (list of cocited papers for each node and the count of the list)
  6. cocitation1.csv - common successors (number of common papers which cite the nodes, for each pair of nodes)
  7. katz (1).csv - Katz centrality (node with highest katz score for each centrality)
  8. pr_dc_ec.csv - Page rank, Degree centrality and Eigen Centrality (Nodes with highest score for these measures, for each community)
  9. prestige.csv - Degree, Proximity and Rank Prestige (for 10 communities)
  10. re_cc_kc.csv - Reciprocal of Eccentrity, Closeness and Katz centrality (27 communities)
  11. source_nodes.csv - Source node of each community
  12. voterank.csv - Voterank Algorithm (node with highest score for each community)