Draft README.md

README.md

@@ -1,3 +1,129 @@
 ---
 license: odc-by
+task_categories:
+- text-classification
+language:
+- en
+tags:
+- tense-prediction
+- finance
+- nlp
+size_categories:
+- 100K<n<1M
 ---
+
+<p align="center">
+  <img src="https://github.com/NosibleAI/nosible-py/blob/main/docs/_static/readme.png?raw=true"/>
+<p>
+
+# NOSIBLE Predictive Dataset
+
+## Changelog
+- **v1.0.0:** Initial version
+
+## Who is NOSIBLE?
+
+NOSIBLE is a web-scale vertical search engine. Our worldwide media surveillance products help companies build AI systems that see every worldwide event and act with complete situational awareness. In short, we help companies know everything, all the time. The financial institutions we work with rely on us to deliver media intelligence from every country in every language in real-time. Shortcomings in existing financial datasets and financial models are what inspired us to release this dataset and related models.
+
+- [NOSIBLE Financial Sentiment v1.1 Base](https://huggingface.co/NOSIBLE/financial-sentiment-v1.1-base)
+- [NOSIBLE Forward Looking v1.1 Base](https://huggingface.co/NOSIBLE/forward-looking-v1.1-base)
+- **[NOSIBLE Prediction v1.1 Base](https://huggingface.co/NOSIBLE/prediction-v1.1-base)**
+
+## What is it?
+
+The NOSIBLE Forward Looking Dataset is an open collection of **100,000** cleaned, deduplicated, and labeled financial news text samples. The data was extracted directly from the NOSIBLE search engine in response to real queries asked by real financial institutions. Each sample has been labeled as **predictive** or **not-predictive**, using a sophisticated labeling pipeline (more information coming soon, and a brief explanation below).
+
+## How to use it
+Using the [HuggingFace datasets library](https://huggingface.co/docs/datasets/):
+
+You can install it with `pip install datasets`, and must login using e.g. `hf auth login` to access this dataset.
+
+```python
+from datasets import load_dataset
+
+dataset = load_dataset("NOSIBLE/predictive")
+
+print(dataset)
+
+# DatasetDict({
+#     train: Dataset({
+#         features: ['text', 'label', 'netloc', 'url'],
+#         num_rows: 100000
+#     })
+# })
+
+# What's next?
+#     Train your model 🤖
+#     Profit 💰
+```
+
+## Dataset Structure
+
+### Data Instances
+
+The following is an example sample from the dataset:
+
+```json
+{
+    "text": "Offshore staff HOUSTON \u2013 VAALCO Energy is looking to bring in a floating storage and offloading (FSO) unit at the Etame Marin oil field offshore Gabon. The company has signed a non-binding letter of intent with Omni Offshore Terminals to supply and operate the vessel at Etame for up to 11 years, following the expiry of the current FPSO Petr\u00f3leo Nautipa contract with BW Offshore in September 2022. Omni has provided a preliminary proposal for leasing and operating the FSO, which could reduce VAALCO's operating costs by 15-25%, compared with the current FPSO contract during the term of the proposed agreement. Maintaining the current FPSO beyond its contract or transitioning to a different FPSO, VAALCO added, would require substantial capex investments.",
+    "label": "predictive",
+    "netloc": "www.offshore-mag.com",
+    "url": "https://www.offshore-mag.com/rigs-vessels/article/14202267/vaalco-contemplating-switch-to-fso-at-etame-offshore-gabon"
+}
+```
+
+### Data Fields
+
+- `text` (string): A text chunk from a document.
+- `label` (string): The causal structure label of the text, sourced from LLMs and refined with active learning (an iterative relabeling process).
+- `netloc` (string): The network location (domain) of the document.
+- `url` (string): The URL of the document.
+
+## Dataset creation
+
+### Data source
+The dataset was sampled from NOSIBLE datafeeds, which provides web-scale surveillance data to customers. Samples consist of top-ranked search results from the NOSIBLE search engine in response to safe, curated queries. All data is sourced exclusively from the public web.
+
+### Relabeling algorithm
+The dataset's label field was annotated by LLMs and refined using an active learning algorithm called relabeling.
+
+The algorithm outline is as follows:
+
+1. Hand label a candidate set of ~200 samples to use as a test bed to refine the prompt used by the LLM labelers to classify the text.
+2. Label a set of 100k samples with LLM labelers:
+    - `x-ai/grok-4-fast`
+    - `x-ai/grok-4-fast:thinking`
+    - `google/gemini-2.5-flash`
+    - `openai/gpt-5-nano`
+    - `openai/gpt-4.1-mini`
+    - `openai/gpt-oss-120b`
+    - `meta-llama/llama-4-maverick`
+    - `qwen/qwen3-32b`
+3. Train a linear model on the labels using the majority vote of the LLM labelers.
+4. Iterative relabeling (active learning steps) to improve the label quality:
+    - Evaluate linear model's predictions over the samples.
+    - Find disagreements: samples where the LLM labelers agree on a label, but the model has predicted a different label.
+    - Consult a much larger LLM, the oracle, to evaluate the model's prediction and relabel the sample if it agrees with the LLM labels.
+    - Drop the worst performing LLM labelers from the ensemble.
+    - Repeat the process with the remaining LLM labelers until the number of samples relabeled reaches 0.
+    - Store the refined relabeled dataset.
+5. This is the final dataset used for training the [NOSIBLE Prediction v1.1 Base](https://huggingface.co/NOSIBLE/prediction-v1.1-base) model, which is a finetune.
+
+## Additional information
+
+### License
+The dataset is released under the **Open Data Commons Attribution License (ODC-By) v1.0** [license](https://opendatacommons.org/licenses/by/1-0/).
+
+### Attribution
+- [NOSIBLE Inc](https://www.nosible.com/) Team
+
+**Contributors**
+
+This dataset was developed and maintained by the following team:
+
+* [**Matthew Dicks**](https://www.linkedin.com/in/matthewdicks98/)
+* [**Simon van Dyk**](https://www.linkedin.com/in/simon-van-dyk/)
+* [**Stuart Reid**](https://www.linkedin.com/in/stuartgordonreid/)
+
+## Citations
+Coming soon, we're working on a white paper.