Abstract | ||
---|---|---|
This paper studies the learning problem of ranking when one wishes not just to ac- curately predict pairwise ordering but also preserve the magnitude of the preferences or the dierence between ratings, a prob- lem motivated by its key importance in the design of search engines, movie recom- mendation, and other similar ranking sys- tems. We describe and analyze several al- gorithms for this problem and give stabil- ity bounds for their generalization error, ex- tending previously known stability results to non-bipartite ranking and magnitude of preference-preserving algorithms. We also re- port the results of experiments comparing these algorithms on several datasets and com- pare these results with those obtained using an algorithm minimizing the pairwise mis- ranking error and standard regression. |
Year | DOI | Venue |
---|---|---|
2007 | 10.1145/1273496.1273518 | ICML |
Keywords | Field | DocType |
movie recommendation,paper study,stability bound,similar ranking system,stability result,ranking algorithm,search engine,preference-preserving algorithm,generalization error,key importance,pairwise misranking error | Data mining,Magnitude (mathematics),Learning to rank,Ranking SVM,Computer science,Ranking (information retrieval),Artificial intelligence,Pairwise comparison,Search engine,Ranking,Pattern recognition,Regression,Machine learning | Conference |
Citations | PageRank | References |
53 | 1.88 | 9 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Corinna Cortes | 1 | 6574 | 1120.50 |
Mehryar Mohri | 2 | 4502 | 448.21 |
Ashish Rastogi | 3 | 161 | 10.55 |