1st International ICST Conference on Scalable Information Systems

Research Article

A generic search strategy for large-scale real-world networks

  • @INPROCEEDINGS{10.1145/1146847.1146849,
        author={Yuichi  Kurumida and Tsukasa  Ogata and Hirotaka  Ono and Kunihiko  Sadakane and Masafumi  Yamashita},
        title={A generic search strategy for large-scale real-world networks},
        proceedings={1st International ICST Conference on Scalable Information Systems},
        publisher={ACM},
        proceedings_a={INFOSCALE},
        year={2006},
        month={6},
        keywords={},
        doi={10.1145/1146847.1146849}
    }
    
  • Yuichi Kurumida
    Tsukasa Ogata
    Hirotaka Ono
    Kunihiko Sadakane
    Masafumi Yamashita
    Year: 2006
    A generic search strategy for large-scale real-world networks
    INFOSCALE
    ACM
    DOI: 10.1145/1146847.1146849
Yuichi Kurumida1,2,*, Tsukasa Ogata3,*, Hirotaka Ono3,*, Kunihiko Sadakane3,*, Masafumi Yamashita3,*
  • 1: Department of Electrical Engineering and Computer Science, Kyushu University
  • 2: 6-10-1 Hakozaki, Fukuoka, Japan
  • 3: Department of Computer Science and Communication Engineering, Kyushu University, 6-10-1 Hakozaki, Fukuoka, Japan.
*Contact email: kurumida@tcslab.csce.kyushu-u.ac.jp, tsukasa@tcslab.csce.kyushu-u.ac.jp, ono@csce.kyushu-u.ac.jp, sada@csce.kyushu-u.ac.jp, mak@csce.kyushu-u.ac.jp

Abstract

We consider the following situation for a given large-scale network: Starting from an initial node we move to its neighbor node and repeat that until reaching a target node. How fast can we do this without any global topological information? This problem is considered "searching networks", and several approaches have been proposed. In this paper, we present a general framework of search strategies, under which all of these existing approaches can be formalized. Our framework characterizes random search strategies by the following three parameters: memory for previously-visited nodes, look-ahead property and transition probability. Through computational simulations for large-scale networks with small-worldness and scale-freeness, we investigate the relationship between the effect of parameters of the strategies and the coefficients of networks such as the clustering coefficient. The comparison result provides a guideline to obtain good parameters of the strategies according to the diameter and the clustering coefficients of networks.