SetRank: Learning a Permutation-Invariant Ranking Model for Information Retrieval

总结

加序列emb,multi-head self-attention/transformer

细节

当输入list排序变化后,用rank模型输出不变的排序list。multi-head self-attention堆叠解决。

representation-encoding-ranking

SetRank: Learning a Permutation-Invariant Ranking Model for Information Retrieval_第1张图片

先用现有的ranking model做出来init ranking,再multi-head attention做encode,最后fnn做ranking。

交叉熵损失

实验

dataset

  1. Istella LETOR:http://blog.istella.it/istella-learning-to-rank-dataset/
  2. Microsoft LETOR 30K:http://research.microsoft.com/en-us/projects/mslr/
  3. Yahoo! LETOR:http://learningtorankchallenge.yahoo.com

baseline:rankSVM, rankBoost, MART, LambdaMart, DLCM, GSF

评估指标:ndcg@1,3,5,10

你可能感兴趣的:(机器学习,transformer,深度学习,机器学习)