Abstract
The basis of Google’s acclaimed PageRank is an artificial mixing of the Markov chain representing the connectivity structure of the network under study with a maximally connected network where every node is connected to every other node. The rate with which the original network is mixed with the strongly connected one is called damping factor. The choice of the damping factor can influence the ranking of the nodes. As we show in this paper, the ranks of transient nodes, i.e., nodes not belonging to a strongly connected component without outgoing links in the original network, tend to zero as the damping factor increases. In this paper we develop a new methodology for obtaining a meaningful ranking of nodes without having to resort to mixing the network with an artificial one. Our new ranking relies on an adjusted definition of the ergodic projector of the Markov chain representing the original network. We will show how the new ergodic projector leads to a more structural way of ranking (transient) nodes. Numerical examples are provided to illustrate the impact of this new ranking approach.
Similar content being viewed by others
Notes
Positivity of w is needed for the row normalization later on. Possibly a scaling is required to make the weights positive in practical applications.
The term random walk also applies, but since the ranking is inspired by Google’s PageRank we will use their concept of random surfer.
References
Aldous D, Fill J (2002) Reversible Markov Chains and Random Walks on Graphs. online version available at http://www.stat.berkeley.edu/users/aldous/RWG/book.html. Unfinished monograph, recompiled in 2014
Avrachenkov K, Litvak N (2006) The effect of new links on Google Pagerank. Stoch Model 22(2):319–331
Avrachenkov K, Litvak N, Pham KS (2008) A singular perturbation approach for choosing the PageRank damping factor. Internet Math 5(1-2):47–69
Berkhin P (2005) A survey on PageRank computing. Internet Math 2 (1):73–120. doi:http://dx.doi.org/10.1080/15427951.2005.10129098
Berkhout J (2016) Google’s PageRank algorithm for ranking nodes in general networks. In: Proceedings of the 13th international workshop on discrete event systems (WODES 2016), pp 153–158
Berkhout J, Heidergott BF (2014) A series expansion approach to risk analysis of an inventory system with sourcing. In: Proceedings of the 12th international workshop on discrete event systems (WODES 2014), vol 12, pp 510–515
Berkhout J, Heidergott BF (2015) Efficient algorithm for computing the ergodic projector of Markov multi-chains. Procedia Comput Sci 51:1818–1827. doi:10.1016/j.procs.2015.05.403. http://www.sciencedirect.com/science/article/pii/S1877050915012119
Bianchini M, Gori M, Scarselli F (2005) Inside PageRank. ACM Trans Internet Technol 5 (1):92–128. doi:10.1145/1052934.1052938. http://delivery.acm.org/10.1145/1060000/1052938/p92-bianchini.pdf?ip=208.165.55.137&id=1052938&acc=ACTIVESERVICE&key=842BC1E250410AEB.FDB887D4E02C11F2.4D4702B0C3E38B35.4D4702B0C3E38B35&CFID=710709363&CFTOKEN=16287877&__acm__=1441563203_9363b1527fecd54000
Boldi P, Santini M, Vigna S (2005) PageRank as a function of the damping factor. In: Proceedings of the 14th international conference on World Wide Web - WWW ’05 pp 557. doi:10.1145/1060745.1060827. http://portal.acm.org/citation.cfm?doid=1060745.1060827
Boldi P, Santini M, Vigna S (2009) PageRank: functional dependencies. ACM Trans Inf Syst 27(4):1–23. doi:10.1145/1629096.1629097
Boldi P, Vigna S (2014) Axioms for centrality. Internet Math 10(3-4):222–262
Bressan M, Peserico E (2010) Choose the damping, choose the ranking? J Discret Algorithms 8(2):199–213. doi:10.1016/j.jda.2009.11.001
Brin S, Page L (1998) The anatomy of a large-scale hypertextual Web search engine. Computer Networks and ISDN Systems 30(1-7):107–117. http://www.sciencedirect.com/science/article/pii/S016975529800110X
Broder A, Kumar R, Maghoul F, Raghavan P, Rajagopalan S, Stata R, Tomkins A, Wiener J (2000) Graph structure in the Web. Comput Netw 33 (1):309–320. doi:10.1016/S1389-1286(00)00083-9
Franceschet M (2011) PageRank: standing on the shoulders of giants. Commun ACM 54(6):92–101. http://dl.acm.org/citation.cfm?id=1953146
Fu HH, Lin DKJ, Tsai HT (2006) Damping factor in Google page ranking. Appl Stoch Model Bus Ind 22(5-6):431–444. doi:10.1002/asmb.656
Gleich DF (2015) PageRank beyond the web. SIAM Rev 57(3):321–363. doi:10.1137/140976649. arXiv:1407.5107
Heidergott BF, Hordijk A, van Uitert M (2007) Series expansions for finite-state Markov chains. Probab Eng Informational Sci 21(3):381–400. doi:10.1017/S0269964807000034. http://dl.acm.org/citation.cfm?id=1291117.1291120
Kartashov NV (1996) Strong stable markov chains. TBIMC Scientific Publishers, VSP
Kemeny JG, Snell JL (1976) Finite Markov chains: With a New Appendix “Generalization of a Fundamental Matrix. Springer, New York. http://books.google.nl/books/about/Finite_markov_chains.html?id=WORLAAAAMAAJ&pgis=1
Langville AN, Meyer CD (2003) Fiddling with PageRank. CRSC- TR03-34. North Carolina State University, Raleigh, NC. http://www.ncsu.edu/crsc/reports/reports03.html Tech. rep.
Langville AN, Meyer CD (2004) Deeper inside PageRank. Internet Math 1(3):335–380. doi:10.1080/15427951.2004.10129091
Langville AN, Meyer CD (2005) A survey of eigenvector methods for web information retrieval. SIAM Rev 47(1):135–161. doi:10.1137/S0036144503424786
Langville AN, Meyer CD (2011) Google’s pagerank and beyond: The science of search engine rankings. Princeton University Press
Langville AN, Meyer CD (2012) Do the math: why no ranking system is no. 1. Sci Am 307(1)
Leskovec J, Krevl A (2014). SNAP Datasets: Stanford Large Network Dataset Collection, Webpage:https://snap.stanford.edu/data/wiki-Vote.html
Meyer CD (1975) The role of the group generalized inverse in the theory of finite Markov chains. SIAM Rev 17(3):443–464. doi:10.1137/1017044
Meyer CD (1982) Analysis of finite Markov chains by group inversion techniques. In: Campbell SL (ed) Recent applications of generalized inverses, research notes in mathematics, vol 66. Pitman, Boston, London, pp 50–81
Tarjan RE (1972) Depth-first search and linear graph algorithms. SIAM J Comput 1(2):146–160
Xia L, Glynn PW (2016) A generalized fundamental matrix for computing fundamental quantities of Markov systems. arXiv:1604.04343v2 [math.OC]. 1604.04343
Author information
Authors and Affiliations
Corresponding author
Additional information
This article is part of the Topical Collection: Special Issue on Performance Analysis and Optimization of Discrete Event Systems
Guest Editors: Christos G. Cassandras and Alessandro Giua
Part of this work was done while Joost Berkhout was with Vrije Universiteit Amsterdam
Rights and permissions
About this article
Cite this article
Berkhout, J., Heidergott, B.F. Ranking nodes in general networks: a Markov multi-chain approach. Discrete Event Dyn Syst 28, 3–33 (2018). https://doi.org/10.1007/s10626-017-0248-7
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10626-017-0248-7