cash-app-inc/open-sea-rlr

---
fontsize: 10pt
geometry: margin=0cm
output: pdf_document
license: cc-by-4.0
---
# Open Sea for Recent Link Regression
## Summary
1. **Name:** `open-sea-rlr`
2. **Description:** This dataset involves a Non-Fungible Token (NFT) transaction network between the users of [Open Sea](https://opensea.io/). Vertices represent unique sellers and buyers of NFTs. Each transaction is characterized by a timestamp and a feature vector. Features include binary representations of categorical variables, cryptocurrency exchange rates, and monetary values. The target value for each transaction is computed as the rate of return on investment, normalized by the original purchase price. This metric reflects the profitability of each transaction.
3. **Task:** The task is predicting the rate of return for an investment (NFT purchase).
4. **Date of Creation:** 01.07.2024
5. **Last Update:** 01.07.2024
6. **Original Sources:** https://huggingface.co/datasets/MLNTeam-Unical/NFT-70M_transactions
7. **Contact Information:** [email]([e-mail])
8. **License:** CC BY 4.0

## Statistics
| Category | Data |
|-------------------------|------------|
| Number of Nodes | 2,601,107 |
| Number of Edges | 25,876,356 |
| Number of Node Features | 0 |
| Number of Edge Features | 86 |
| Number of Timestamps | 7,361,184 |

## Download
1. **Format:** Compressed `data.pt` which involves a python dictionary as follows:
```python
data = {
"node_attr": None,
"edge_index": torch.LongTensor,
"edge_time": torch.FloatTensor,
"edge_attr": torch.FloatTensor,
"edge_label": torch.FloatTensor,
"num_nodes": int
}
```
2. **Size:** 322 MB
3. **Location:** https://huggingface.co/datasets/ca-aird/airtraffic2015/blob/main/data.zip

## Citation
```
@article{,
title={Benchmarking Edge Regression on Temporal Networks},
author={Muberra Ozmen and Florence Regol and Thomas Markovich},
journal={X},
volume={X},
number={X},
pages={X},
year={X},
publisher={X}
}
```

## Preprocessing
Transaction records involve following fields:

| Field | Description | Usage |
|------------------------------------------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------|-----------------------------------|
| `seller_account` | Address of the NFT seller. | Used as source node |
| `winner_account` | Address of the NFT buyer. | Used as destination node |
| `tx_timestamp` | Timestamp of the transaction. | Used as edge time |
| `token` | Token type used to pay the transaction: converted to categorical data with 0/1 indicator. | Used as edge feature |
| `chain` | Blockchain where the transaction occurs: converted to categorical data with 0/1 indicator. | Used as edge feature |
| `token_type` | Schema of the token, i.e., ERC721 or ERC1155: converted to categorical data with 0/1 indicator. | Used as edge feature |
| `asset_contract_type` | Asset typology, i.e., non-fungible or semi-fungible: converted to categorical data with 0/1 indicator. | Used as edge feature |
| `asset_type` | Whether the asset was involved in a simple or bundle transaction: converted to categorical data with 0/1 indicator. | Used as edge feature |
| `to_eth` | Conversion rate to convert tokens into Ethereum at the current timestamp: normalized to [0, 1]. | Used as edge feature |
| `to_usd` | Conversion rate to convert tokens into US dollars (USD) at the current timestamp: normalized to [0, 1]. | Used as edge feature |
| `created_date` | Date of creation of the contract: converted to timestamp. | Used as edge feature |
| `token_id` | ID of the NFT. | |
| `collection_name` | ID for accessing the collection name: `token_id` is unique within the same collection, so these two are used to identify unique item identification `item_id`. | |
| `usd_price` | Price of the transaction expressed in US dollars (USD): used to calculate edge target such that (`usd_price` - NEXT[`usd_gain`]) / `usd_price`. | Used to calculate edge target |
| `usd_gain` | Difference between the price and the fees expressed in US dollars (USD): used to calculate edge target such that (`usd_price` - NEXT[`usd_gain`]) / `usd_price` | Used to calculate edge target |

## Notes
1. **Acknowledgements:**
- All funding was provided by [Block Inc](https://block.xyz).
2. **References:**
- La Cava, L., Costa, D., & Tagarelli, A. (2023). SONAR: Web-based Tool for Multimodal Exploration of Non-Fungible Token Inspiration Networks. In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR '23), 3200–3204. New York, NY, USA: Association for Computing Machinery.
- La Cava, L., Costa, D., & Tagarelli, A. (2023). Visually Wired NFTs: Exploring the Role of Inspiration in Non-Fungible Tokens. arXiv preprint arXiv:2303.17031.
- Costa, D., La Cava, L., & Tagarelli, A. (2023). Show me your NFT and I tell you how it will perform: Multimodal representation learning for NFT selling price prediction. In Proceedings of the ACM Web Conference 2023 (WWW '23), 1875–1885. New York, NY, USA: Association for Computing Machinery.

## Author Statement
I, Muberra Ozmen, declare that I bear full responsibility for the dataset described herein, including its contents and compliance with applicable laws and regulations. By providing access to this dataset, I confirm that it is released under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0). Users are free to:

- **Share** — copy and redistribute the material in any medium or format
- **Adapt** — remix, transform, and build upon the material for any purpose, even commercially

Under the following conditions:

- **Attribution** — appropriate credit must be given to the associated publication indicating if any changes were made. This information should be provided in a manner that is reasonable given the medium, means, and context in which the dataset is shared.

For any use or redistribution of the dataset not permitted under this license, explicit permission from the dataset's creator is required.

The dataset will be hosted on a secure platform that ensures continuous access to the data. We have chosen [Hugging Face](https://huggingface.co/) for its robust infrastructure and capability to handle large datasets. Access to the dataset will be facilitated through a curated interface, providing users with efficient search and retrieval functionalities.

**Licensing:**
The dataset is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0), allowing users to freely use, share, and adapt the material, provided appropriate credit is given to the dataset's creator.

**Maintenance:**
Regular maintenance of the dataset and its hosting platform will be conducted to ensure data integrity, security, and accessibility. Updates to the dataset, if any, will be promptly integrated into the platform to reflect the most current information available.