Explainable ranking function using Evidence based learning

This is the code repository for the explainable ranking model based on evidence-based learning approach.

Description: We learn an explainable ranknig function from search logs based on probabilistic models and learning to rank methods. This approach is also used to generate explainable content-matching features that are added to LTR models to improve performance and explainability.Report

Contents

Code/ : learning explainable features from search logs to add to ranking function

RankLib/ : for ranking using learning to rank methods including LAMBDMART

Results/ : contains results/performance of the Evidence based learning(EBL) approach, LTR baseline model, LTR+EBL model in terming of NDCG@k, precision@k and recall@k.

License

By using this source code you agree to the license described in https://github.com/sahitilucky/Precision_recall_effort_query_formulation_model/blob/master/LICENSE.md.