Understanding domain "relevance" in web search

This paper presents results of a study that analyses people’s behavior in accessing dierent domains (e.g., images, movies, blogs, etc.) in web search. Our results show that the information seeking process of web users involves accessing information from various domains, which suggests the need to provide results from various domains in an aggregated manner. This study also indicates the existence of associations between query categories and domains, which suggests that it is important (and possible) to select domains that are \relevant" to the query.