---
license: cc-by-4.0
tags:
- dataset
- news-media
- pointcloud-text
---

# 📊 News Media Pointcloud Text preparation notes

## Overview

Preparation notes and schema examples for **News Media** tasks using **Pointcloud Text** data. Full source material is intentionally not bundled, so provenance and licensing remain explicit.

Comments and corrections can be left in the discussion tab.

---

## Intended Use

The repository is suitable for testing the data pipeline, adapting it to a documented source, and preparing controlled research splits. Release status: **metadata sample; full source data not bundled**. The sample is for schema inspection only and should not be reported as a full training corpus.

---

## Environment

| Item | Value |
|---|---|
| Python | >= 3.9 |
| torch | 2.x |
| safetensors | any recent |

---

## Included Material

1. `dataloader.py` — loading, cleaning and split preparation.
2. `dataset_infos.json` — schema and split metadata.
3. `metadata_sample.jsonl` — readable records for schema checks.

---

## Processing Choices

| Item | Value |
|---|---|
| Storage format | webdataset |
| Preprocessing | standard |
| Augmentation | none |
| Split strategy | leave one out |
| Sampling | stratified |
| Quality checks | adaptive |
| Labeling | pseudo label |

---

## Data Sources

Source material is referenced but not bundled. Record the source version, retrieval date and any filtering rule so the same rows can be reproduced later.

---

## Risks And Limitations

External sources may be unrepresentative, mis-licensed, or contain sensitive fields. Source review, privacy checks and bias analysis are the user's responsibility before training or redistribution.

---

## Files

1. `dataloader.py` — primary artifact
2. `README.md` — this documentation
3. `dataset_infos.json` — schema metadata
4. `metadata_sample.jsonl` — schema sample
5. `examples/sample_usage.py` — usage sketch

---

## License

Distributed under the **cc-by-4.0** license. External datasets keep their own terms — check them separately.