Modeling Label Ambiguity for Neural List-Wise Learning to Rank
This addresses a specific issue in information retrieval for ranking tasks, offering an incremental improvement over existing list-wise methods.
The paper tackled the problem of ignoring label ambiguity in list-wise learning to rank, where multiple documents share the same relevance label, by proposing a novel sampling technique for computing a list-wise loss that accounts for this ambiguity. The result showed that the method generalizes better and significantly outperforms strong baselines like ListNet and ListMLE on validation and test sets.
List-wise learning to rank methods are considered to be the state-of-the-art. One of the major problems with these methods is that the ambiguous nature of relevance labels in learning to rank data is ignored. Ambiguity of relevance labels refers to the phenomenon that multiple documents may be assigned the same relevance label for a given query, so that no preference order should be learned for those documents. In this paper we propose a novel sampling technique for computing a list-wise loss that can take into account this ambiguity. We show the effectiveness of the proposed method by training a 3-layer deep neural network. We compare our new loss function to two strong baselines: ListNet and ListMLE. We show that our method generalizes better and significantly outperforms other methods on the validation and test sets.