README.md

---
task_categories:
- other
tags:
- recommendation-system
- poi-recommendation
- mobility
- travel
size_categories:
- 1B<n<10B
configs:
- config_name: default
  data_files:
  - split: interaction
    path: interaction_1.csv
  - split: user_info
    path: user_info.csv
  - split: poi_info
    path: poi_info.csv
---

**All data has been uploaded. Please note that the current POI dataset does not include coordinate information. We will be updating it as soon as possible, so please stay tuned. If you do not need the coordinate information for the POIs, please disregard this message.**

# IntTravel: A Real-World Dataset and Generative Framework for Integrated Multi-Task Travel Recommendation

[Paper](https://huggingface.co/papers/2602.11664) | [GitHub](https://github.com/AMAP-ML/IntTravel)

IntTravel is the first large-scale public dataset for integrated travel recommendation, including **4.1 billion interactions from 163 million users with 7.3 million POIs**. Built upon this dataset, the authors introduce an end-to-end, decoder-only generative framework for multi-task recommendation. It incorporates information preservation, selection, and factorization to balance task collaboration with specialized differentiation.

## Dataset Structure

The IntTravel dataset provides a comprehensive view of user journeys, including "when to depart", "how to travel", "where to go", and "what needs arise via the route".

### Information of POIs
The dataset contains **7,291,872** POIs distributed across several major cities in China. The POI data is stored in the `poi_info.csv` file. The `poi_info_groupby_gid.csv` file is produced based on `poi_info.csv` for the convenience of geographic negative sampling.

| Field | Description |
|:---|:---|
| POI ID | A unique identifier for each Point of Interest. |
| Normalized score | A 0-1 score reflecting the overall popularity of the POI. |
| Geographic ID | Identifier for the POI's geographic block. Same GIDs indicate geographical proximity. |
| Category ID | A numerical identifier for the Point of Interest's category. |
| Administrative Region ID | The identifier for the administrative region of the POI. |
| Coordinates | The spatial coordinates of the POI on a 2D plane. |

### User Profiles
The dataset contains **162,815,861** users. The user data is stored in the `user_info.csv` file.

| Field | Description |
|:---|:---|
| User ID | A unique identifier assigned to each user. |
| Profile Feature 1-6 | Six profile features describing the user. |

### User Interactions
The dataset includes **4,129,827,011** user interaction events. Due to the large volume of data, the user interaction data has been split into 100 files, ranging from `interaction_1.csv` to `interaction_100.csv`, with each file containing approximately 40 million records.

| Field | Description |
|:---|:---|
| User ID | A unique identifier for the user who performed the interaction. |
| Timestamp | The time of the user interaction, recorded in milliseconds. |
| Action Type | A numerical ID representing the type of user behavior (e.g., click). |
| POI ID | The identifier of the Point of Interest involved in the interaction. |
| Geographic ID | The geographic block ID where the user was during the interaction. |
| Administrative Region ID | The administrative region ID where the user was during the interaction. |
| Weather | A numerical ID representing the weather condition during the interaction. |
| Travel Mode | A numerical ID for the user's chosen travel mode. |
| Via POI ID | The identifier for a way-point POI added by the user. |

## Citation
If you use this dataset in your research, please cite the following paper:
```bibtex
@article{yan2025inttravel,
  title={IntTravel: A Real-World Dataset and Generative Framework for Integrated Multi-Task Travel Recommendation},
  author={Huimin Yan and Longfei Xu and Junjie Sun and Zheng Liu and Wei Luo and Kaikui Liu and Xiangxiang Chu},
  journal={arXiv preprint arXiv:2602.11664},
  year={2025}
}
```