Learning to rank in heterogeneous network

Zi Yang,Jie Tang,Juanzi Li
DOI: https://doi.org/10.3969/j.issn.2095-2783.2011.04.005
2011-01-01
Abstract:Massive heterogeneous resources have been currently available online,which brings richer information,along with more challenges.This paper concentrates on the issues of "learning to rank" in heterogeneous networks.First of all,the paper describes the investigation on topic-level random walk.We propose a three-step approach,especially focus on topic modeling of documents and the query and calculating a topic-level ranking score.Besides,we propose a general framework for heterogeneous cross-domain ranking,which simultaneously models the correlation between the source domain and the target domain,as well as learns the ranking models.We also develop an efficient EM-style solution,and discuss the generalized bound.The experiments show our proposed methods outperform the baseline methods.Finally,the thesis introduces the concept of specific expert search,which can be solved by the heterogeneous cross-domain ranking algorithm.
What problem does this paper attempt to address?