Iterative learning to rank from explicit relevance feedback
Mateus M. Pereira, Elham Etemad, Fernando V. Paulovich · 2020
Interactive information retrieval (IIR) models consider the implicit or explicit user feedback in order to understand their intent and improve the quality of retrieved documents. We are proposing an IIR model that combines explicit relevance feedback and learning to rank techniques to improve the quality of retrieved documents. Besides evaluating the proposed method in general information retrieval cases using LETOR datasets, we have applied it to the special case of Community Question Answering (CQA) systems using SemEval challenge data. The proposed method outperforms other existing learning to rank techniques on most of these datasets.