DNS usage mining and its two applications

Author

Wu, Jun ; Li, Xiaodong ; Wang, Xin ; Yan, Baoping

Author_Institution

Grad. Sch. of Chinese Acad. of Sci., Beijing, China

fYear

2011

fDate

26-28 Sept. 2011

Firstpage

54

Lastpage

60

Abstract

The principal goal of DNS usage mining is the discovery and analysis of patterns in the query behavior of DNS users. In this paper, we develop a unified framework for DNS usage mining based on Clustering analysis of co-occurrence data derived from DNS server query data. Through transforming the raw query data into co-occurrence matrix, user transaction clustering can be applied to discover the user groups according to their similar query behaviors. Using the aggregate usage profile that represents a user cluster and suitable similarity measure, a specific approach for a domain name recommendation engine is shown. For identifying the latent purpose of a domain name, Probabilistic Latent Semantic Analysis (PLSA) is used, which can automatically discover hidden semantic relationships between users and domain names. We demonstrate the effectiveness of our approaches through experiments performed on real-world data sets.

Keywords

data analysis; data mining; pattern clustering; query processing; recommender systems; statistical analysis; DNS usage mining; aggregate usage profile; clustering analysis; co-occurrence data analysis; domain name recommendation engine; domain name service; pattern analysis; pattern discovery; probabilistic latent semantic analysis; query behavior; user transaction clustering; Aggregates; Context; Current measurement; Data mining; Engines; Internet; Vectors;

fLanguage

English

Publisher

ieee

Conference_Titel

Digital Information Management (ICDIM), 2011 Sixth International Conference on

Conference_Location

Melbourn, QLD

ISSN

Pending

Print_ISBN

978-1-4577-1538-9

Type

conf

DOI

10.1109/ICDIM.2011.6093364

Filename

6093364