FULTR: A Large-scale Prior-Posterior Fusion Learning to Rank Dataset and its application for Satisfaction-Oriented Ranking

In this paper, we introduce FULTR–a large-scale, prior-posterior FUsion LTR dataset. FULTR comprises over 224M queries and 683M documents from Baidu Search, combining both: (1) a rich prior-attribute set with detailed textual relevance, authority, recency, and quality features, and (2) a comprehensive posterior-attribute set enriched by user click data, dwell time, and positional information.

Dataset Folder Structure

The dataset is organized as follows:

FULTR/
├── prior/ # prior-attribute set
│ ├── train.gz
│ ├── test.gz
├── posterior/ # posterior-attribute set
│ ├── train/
│ │ ├──part-00005.gz
│ │ ├──part-00006.gz
│ │ ├──...
│ │ ├──part-00105.gz
│ ├── test/
│ │ ├──part-00000.gz
│ │ ├──part-00001.gz
│ │ ├──part-00002.gz
│ │ ├──part-00003.gz
│ │ ├──part-00004.gz

Download the Dataset

You can download the dataset using the link: Download Dataset. Due to legal restrictions with Baidu, we only made part of the training set of the posterior-attribute set available. If you want the full data, please contact the authors and explain the reason for using this dataset.

Quick Start

The code for training models using this dataset will be released soon. Stay tuned for updates!

License

This dataset is licensed under the CC BY-NC 4.0 license. You are free to share and adapt the dataset for non-commercial purposes, but proper attribution is required. For commercial use, please contact the authors.

Contact

For any questions or suggestions, please contact:
EMAIL: yuchenli1230 {at} gmail.com