Exploring and Understanding Web Search Behavior with Human Activities

Nowadays, web search has become an important approach to retrieving information. Incorporating context factors such as device type and location to improve search performance has attracted more and more attentions. However, impacts of offline human activities (i.e., background activity) on web search haven't been fully explored yet, although intuitively, some of them exist. To reduce this gap, we examine user's web search behavior during activities of different themes by studying a 9-day Internet access logs in a large convention center, as well as web search preferences with respect to other factors, such as device type, query language, in these activities. Our results show that, the background activity causes a large proportion of activity theme related queries, while it has little influence on theme irrelevant ones. In addition, even if in the same activity context, differences in query language, device type, device vendor and search engine lead to various changes in web search behavior.