Ranking algorithm based on structured learning
Yiwen Zhang · Computer Engineering and Applications Journal · 2011
For the problem that the model learned from traditional ranking algorithm which converts ranking problem to classification or regression is not accurate,a novel ranking algorithm is proposed.It views the ranking problem as a procedure of structured learning which learns a rank structure from the train set.The algorithm defines a object function of query level, and presents using the cutting plane algorithm to solve the problem that the algorithm has exponential number of constraints. For the sub-problem of finding the most violated constraints,the paper transforms it into a simple sorting in descending order.Experimental results on the benchmark datasets show that the algorithm proposed in this paper is more effective than the traditional ranking algorithm.