alimamaTech/MAC

---
license: apache-2.0
task_categories:
- other
tags:
- advertising
- conversion-rate-prediction
- multi-attribution-learning
---

# MAC: Multi-Attribution BenChmark

[Paper](https://huggingface.co/papers/2603.02184) | [Code](https://github.com/alimama-tech/PyMAL)

Conversion Rate (CVR) prediction is a cornerstone of online advertising systems. However, existing public CVR datasets—such as Criteo and Ali-CCP—provide conversion labels derived from a single attribution mechanism, severely limiting research into more holistic modeling paradigms. To bridge this gap, we introduce **MAC (Multi-Attribution BenChmark)**, the first public CVR dataset featuring labels from multiple attribution mechanisms.

## Overview
MAC is the first benchmark featuring labels from multiple attribution mechanisms, specifically designed to foster research in Multi-Attribution Learning (MAL). By learning from conversion labels yielded by multiple attribution mechanisms, models can obtain a more comprehensive and robust understanding of touchpoint value.

Along with the dataset, the authors provide **PyMAL**, an open-source library covering a wide array of baseline methods (such as MMoE, PLE, and MoAE) for industrial-scale CVR prediction.

## Dataset Structure
The files in this repository are organized as follows:
- `train/`: Training data files.
- `test/`: Test data files.
- `vocabs/`: ID mappings and vocabulary files for features.

## Usage
To download the dataset directly via `git clone`:
```bash
git clone https://huggingface.co/datasets/alimamaTech/MAC data
```

## Citation
```bibtex
@misc{wu2026macconversionrateprediction,
title={MAC: A Conversion Rate Prediction Benchmark Featuring Labels Under Multiple Attribution Mechanisms},
author={Jinqi Wu and Sishuo Chen and Zhangming Chan and Bird Bai and Lei Zhang and Sheng Chen and Chenghuan Hou and Xiang-Rong Sheng and Han Zhu and Jian Xu and Bo Zheng and Chaoyou Fu},
year={2026},
eprint={2603.02184},
archivePrefix={arXiv},
primaryClass={cs.LG},
url={https://arxiv.org/abs/2603.02184},
}
```