¹School of Computer Science and Technology, Dalian University of Technology, Dalian, China
²School of Software, Dalian University of Technology, Dalian, China

---

Abstract—Link weighting is crucial for performance of linkbased web page ranking algorithms. Typically, a link is viewed as recommendation between pages, which is an unquantifiable term and existing approaches lack physical interpretations. In this paper, we view a link as the attractive force between pages, and map concepts of web (in/out degree, content similarity, etc.) to those of physics (mass, distance, attractive force). Inspired by Reilly’s Law of Retail Gravitation, we propose a gravitation-like model for calculating the attractive force. We then implement a instance of our algorithm framework by taking some features of web pages into consideration. Experimental results show that this instance outperforms other typical algorithms (HITS, Randomized-HITS, and SALSA) with higher precision, better resistibility of TKC effect and no need of filtering intra-domain links.

Index Terms—Link Analysis, Attractive Force Model, Link Weight, Web Information Retrieval.

[PDF]

Cite: Xinyue Liu, Hongfei Lin, Liguo Zhang, "An Attractive Force Model for Weighting Links in Query-Dependant Web Page Ranking," Journal of Computers vol. 7, no. 1, pp. 124-129, 2012.