Personalization poses new challenges to digital library. How to describe users’ preferences and how to support preference queries in digital libraries are among the top demanding tasks. A strict partial order preference model has been proposed, and a series of preference generation methods have already been developed for relational data. However, the semi-structured data stored in digital libraries make it more complex to model users’ preferences than its counterpart in relational databases. The partial order preference model cannot suffice to support such complex preferences. A new ontology based preference model is proposed in this paper to overcome this difficulty. This model describes the documents along with the preferences regarding the documents in the digital libraries using ontology, which allows both the structure and the semantic of a user’s preference to be fully represented. A set of complex preference operations have also been provided in the model to support the personalized query and recommendation efficiently.