Add dataset card with Apache-2.0 license

README.md

@@ -0,0 +1,58 @@
+---
+language:
+- en
+license: apache-2.0
+size_categories:
+- 100K<n<1M
+task_categories:
+- text-generation
+tags:
+- medical
+- dialogue
+- doctor-patient
+- healthcare
+- openmed
+---
+
+# MedDialog
+
+A medical dialogue dataset for training models to generate helpful doctor responses to patient queries.
+
+## Dataset Description
+
+- **Source**: [ruslanmv/ai-medical-chatbot](https://huggingface.co/datasets/ruslanmv/ai-medical-chatbot)
+- **License**: Apache-2.0 (inherited from source dataset)
+- **Size**: ~252k patient-doctor dialogue pairs
+- **Language**: English
+- **Splits**: Train (227k) / Validation (25.2k)
+
+## Dataset Structure
+
+Each example contains:
+- **patient_message**: The patient's description of symptoms or medical question
+- **doctor_response**: The doctor's response (ground truth)
+- **dialogue_context**: Additional context (empty for single-turn dialogues)
+
+## Source and License
+
+This dataset is derived from [ruslanmv/ai-medical-chatbot](https://huggingface.co/datasets/ruslanmv/ai-medical-chatbot), 
+which is released under the [Apache License 2.0](https://www.apache.org/licenses/LICENSE-2.0).
+
+Processing applied:
+- Combined Description and Patient fields into patient_message
+- Filtered out very short or redirect-only responses
+- Truncated messages over 2500 characters
+- 90/10 train/validation split with seed 42
+
+## Usage
+
+
+
+## Intended Use
+
+This dataset is intended for training and evaluating medical dialogue systems in research settings. 
+It should not be used as a substitute for professional medical advice.
+
+## Part of OpenMed
+
+This dataset is part of the [OpenMed](https://huggingface.co/OpenMed) collection of medical NLP resources.