An Algorithm of Street-level Landmark Obtaining Based on Yellow Pages

Abstract

Street-level landmarks are the important foundation for achieving the high-precision geolocation of target IPs. Considering that yellow pages contain a large number of Web and Email domain names corresponding to institutions; the content is stable; and the format is fixed, this paper proposes a street-level landmark obtaining algorithm based on yellow pages. The domain names of institutions in yellow pages are extracted by using regular expression, and the corresponding IPs are parsed. Landmarks are screened according to whether an IP attribution is consistent with the cities where all possible corresponding institutions are located. By using the SLG geolocation algorithm, the landmarks with a geolocation error falling within the evaluation threshold are rated as reliable landmarks. The experimental results show that the proposed algorithm can effectively correct the mis-deletion and mis-evaluation of some landmarks by the existing typical landmark obtaining algorithm: based on 10 Chinese yellow pages (about 2 million institutions in 5 cities) and 3 American yellow pages (about 1 million institutions in 3 cities), a total of 55,960 reliable street-level landmarks for Web and Email servers are obtained. Among the 346,753 Web server IP evaluations, 48,361 landmarks are revised and 40,753 reliable street-level landmarks are augmented compared with the Web-Based landmark obtaining algorithm.