An investigation of cybercrime-related online search behavior vs general search behavior

Information seeking through search engines is one of the most common modes of accessing information over the internet. The purpose of the paper is to understand internet information foraging and user search behavior based on user search sessions. We characterize a search session by the number of the queries and the number of the clicks. A set of search logs from a large search engine is used in the study for the analysis. We identify user sessions using hierarchical agglomerative clustering. Based on information foraging theory, we propose a model that yields the probability distribution for the number of queries and the number of clicks in a search session. The model assumes that users make a sequence of decisions to proceed from one query to another query, continuing as long as the expected value of continuing search exceeds some threshold. The relationship between the number of queries and the number of clicks is examined, and it shows a strong linear relationship. We compare characteristics between the generally sampled search sessions and cybercrime-related search sessions. The results indicate strong regularities in user search sessions, and those two types of sessions show different patterns.