Information Technology Reference
In this chapter, we have introduced some transfer ranking methods. As for the future
work on transfer ranking, the following issues may need to be investigated.
The effectiveness of transfer learning heavily replies on the relation between the
source and target tasks. Researchers usually use the term “relatedness” to describe
the relation. In the literature, there are some meaningful definitions for the relat-
edness of classification tasks, however, it is still unclear how the relatedness for
ranking should be defined.
As can be seen, the existing methods on transfer ranking, no matter feature level
or instance level, are pretty much similar to the transfer learning methods for clas-
sification. It is not clear what kind of impact that the differences between ranking
and classification would make on the problem. It is expected that some novel
transfer learning methods, which are tailored more towards the unique properties
of ranking, may lead to better results.
In fact, these different levels of transfer learning are complementary, how to ef-
fectively combine them into one generic framework is an open question.
Transfer learning is highly related to multi-task learning. Therefore, in addition to
studying transfer ranking, it makes sense to further look at the topics like multi-
task ranking. But again, it is a key problem whether one can effectively leverage
the unique properties of ranking when moving forward in this direction.
1. Chen, D., Yan, J., Xiong, Y., Xue, G.R., Wang, G., Chen, Z.: Knowledge transfer for cross
domain learning to rank. Information Retrieval Journal. Special Issue on Learning to Rank
13(3), doi 10.1007/s10791-009-9111-2 (2010)
2. Herbrich, R., Obermayer, K., Graepel, T.: Large margin rank boundaries for ordinal regression.
In: Advances in Large Margin Classifiers, pp. 115-132 (2000)
3. Joachims, T.: Optimizing search engines using clickthrough data. In: Proceedings of the 8th
ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD
2002), pp. 133-142 (2002)
4. Wu, Q., Burges, C., Svore, K., Gao, J.: Ranking, boosting, and model adaptation. Information
Retrieval Journal. Special Issue on Learning to Rank 13(3), doi 10.1007/s10791-009-9112-1