Recommender system is a focus in the age of information explosion. In this study, with the benefit of social networking service, we propose a User-Centric Integrated Recommendation Model based on combining of users' individualities and commonalities, in which users' interests are focused and their transitions are traced by analyzing users' information access behaviors and histories, and then a sequence of information seeking actions are recommended to target users through dectecting the transitions of their interests focus by interaction of users and the system, and extracting successful experience from a reference user group, in which the reference users are similar to the target users. A set of bookmark tags are used to describe relations of Web pages. The pages accessed by users are classified by the bookmark tags, and grouped into two categories of individual and common interests and their sub-categories. The individual interests are divided into three types: strong interest, weak interest and uncertain interest. The common interests are divided into popular interest, public interest and private interest. In this paper, in addition to describing definitions and measures, we present a mechanism of inferring interest focus and show the system architecture. Finally, the conclusion and further work are introduced.