Create README.md

data/fnxl/README.md

@@ -0,0 +1,38 @@
+# Dataset Card for Dataset Name
+
+The Financial Numerical Extreme Labelling (FNXL) dataset contains over 142k annotated numerals from U.S. SEC 10-K filings, each tagged with one of 2,794 US-GAAP XBRL labels. It is designed to facilitate research in large-scale, fine-grained financial numeral labelling, particularly under an extreme classification setting.
+
+## Dataset Details
+
+### Dataset Description
+
+- **Curated by:** Soumya Sharma, Subhendu Khatuya, Manjunath Hegde, Afreen Shaikh, Koustuv Dasgupta, Pawan Goyal, Niloy Ganguly
+- **Language(s) (NLP):** English (financial domain)
+- **License:** Research-only use
+- **Release Date:** 2023
+
+### Dataset Sources
+
+- **Repository:** [GitHub – FNXL](https://github.com/soummyaah/FNXL)  
+- **Paper:** [arXiv:2306.03723](https://arxiv.org/abs/2306.03723)  
+- **Data Source:** Publicly available U.S. SEC 10-K filings (2019–2021) for 2,339 companies  
+
+
+## Uses
+
+### Direct Use
+
+The dataset can be used for:
+
+- Extreme multi-label classification (assigning US-GAAP XBRL labels to numerals in financial text)  
+- Named entity recognition (NER) with a large label space  
+- Financial document understanding and automation of annotation workflows  
+- Zero-shot label generalization experiments  
+
+### Out-of-Scope Use
+
+The dataset is **not** intended for:
+
+- Predicting financial performance or making investment decisions  
+- Applications involving private, non-public financial documents  
+- Direct commercial use without explicit permission