A 'Pointwise-Query, Listwise-Document' based Query Performance Prediction Approach
Suchana Datta, Sean MacAvaney, Debasis Ganguly, Derek Greene
Abstract
The task of Query Performance Prediction (QPP) in Information Retrieval (IR) involves predicting the relative effectiveness of a search system for a given input query. Supervised approaches for QPP, such as NeuralQPP are often trained on pairs of queries to capture their relative retrieval performance. However, pointwise approaches, such as the recently proposed BERT-QPP, are generally preferable for efficiency reasons. In this paper, we propose a novel end-to-end neural cross-encoder-based approach that is trained pointwise on individual queries, but listwise over the top ranked documents (split into chunks). In contrast to prior work, the network is then trained to predict the number of relevant documents in each chunk for a given query. Our method is thus a split-n-merge technique that instead of predicting the likely number of relevant documents in the top-k, rather predicts the number of relevant documents for each fixed chunk size p(p<k) and then aggregates them for QPP on top-k. Experiments demonstrate that our method is significantly more effective than other supervised and unsupervised QPP approaches yielding improvements of up to 30% on the TREC-DL'20 dataset and by nearly 9% for the MS MARCO Dev set.