Graph-based ranking model for object-level search
Enhong Chen · Journal of Shandong University · 2009
This paper proposes a novel domain-independent object-level ranking model. Given a set of objects and their relationships,this model provides a ranked list of objects based on their relevance to multiple query objects supplied by the user. We present a multi-plane object relationship graph to describe the space of object-level search,an algorithm for evaluating the popularity values of objects based on the object relationship graph,and an algorithm for evaluating the relevance ratings of objects based on the query object as well as merging multiple query objects. The effectiveness of this model is experimentally verified on the ACM data set. This model provides a better paper recommendation performance than PaperRank. This model also outperforms PaperRank on merging the relevance ratings of multiple query objects into a single vector.