Datasets:

datajuicer
/

RealMedConv

The RealMedConv dataset consists of anonymized, real-world dialogues between licensed pharmacists and users seeking over-the-counter (OTC) medication advice. Each conversation is goal-oriented: the pharmacist gathers sufficient symptom information to provide an online and appropriate recommendation. Dialogues are typically concise, spanning 3–5 turns, reflecting the efficient and expert-driven nature of professional medical consultations.

This dataset originates from real user-pharmacist interactions on an Alibaba Group service platform. To ensure strict compliance with data privacy and security standards, the raw logs have undergone rigorous processing—including de-identification, filtering, translation, and controlled perturbation—so that no personally identifiable information (PII) or sensitive content remains. The released version poses no privacy leakage or security risks.

Only a small subset is being open-sourced to support academic research, particularly in the area of goal-directed, proactive dialogue systems for healthcare advisory tasks.

Intended Uses & Limitations

Intended Uses

Research on task-oriented dialogue systems, especially those requiring information-gathering strategies and online recommendation logic.
Development and evaluation of proactive conversational agents in health-related contexts.
Benchmarking models on symptom inquiry efficiency, user intent understanding, and risk-aware response generation.

Limitations

Not for clinical use: This dataset is strictly for non-clinical, academic research. It does not constitute medical advice, nor should it be used to train systems intended for real-world diagnosis or treatment.
Limited disease scope: Only conversations related to 20 common, non-severe conditions (selected from a broader set, with hundreds of conditions) are included.
Machine-Translated English: All dialogues were originally in Chinese and have been machine-translated into English; minor stylistic or semantic artifacts may exist.
Actual medication names are withheld: The <med_search> token indicates where a recommendation would be generated, but specific drug names or dosages are not included in the public release.

License

This dataset is released under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) license.

✅ You are free to:
Share — copy and redistribute the material in any medium or format.
Adapt — remix, transform, and build upon the material.
❌ Under the following terms:
Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made.
NonCommercial — You may not use the material for commercial purposes.
ShareAlike — If you remix, transform, or build upon the material, you must distribute your contributions under the same license (CC BY-NC-SA 4.0 or a compatible one).

🔗 License URL: https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode.en

Note: In addition to this license, users are required to agree to a separate Data Use Commitment Agreement at download time, affirming responsible and ethical usage.

Data Processing & Privacy Safeguards

To protect user privacy and prevent model inversion or re-identification, the following measures were applied:

Strict De-identification: All PII (e.g., names, ID numbers, phone numbers) was removed from raw logs.
Temporal Perturbation: Any residual temporal markers (e.g., birth months) were perturbed to prevent linkage attacks.
Controlled Sampling: Only 2,000 sessions covering 20 representative conditions were selected—insufficient to reconstruct or replicate the full-scale production model.
Translation & Style Normalization: Original Chinese dialogues were translated into English using internal models, with conversational style harmonized for consistency.

Dataset Structure

The dataset is provided in JSONL format with two splits, train.jsonl and test.jsonl.

Each entry contains:

session_id: Anonymized session identifier (integer)
diagn: Broad diagnostic category (e.g., "Upper respiratory tract infection (URTI)")
messages: List of alternating user/assistant utterances, each with:
role: "user" or "assistant"
content: Dialogue text

Example:

{
"session_id": 25312,
"diagn": "Upper respiratory tract infection (URTI)",
"messages": [
{"role": "user", "content": "I've had a sore throat and headache for about a day..."},
{"role": "assistant", "content": "Have you taken any medication before?"},
...
{"role": "assistant", "content": "<med_search>"}
]
}

The <med_search> token marks the point at which the pharmacist would initiate a medication lookup in the original system. No actual drug recommendations are disclosed.

Citation

If you use this dataset in your research, please cite it as follows:

@misc
{realmedconv2025,
author = {DataJuicer Team, Alibaba Group},
title = {{RealMedConv}: A Dataset of Real-World Pharmacist-User Dialogues for Online Medication Advice},
year = {2025},
note = {Released under CC BY-NC-SA 4.0}
}

Acknowledgements

We thank the licensed pharmacists and users whose anonymized interactions made this research possible. This work was conducted by DataJuicer Team, Alibaba Group, in alignment with our commitment to responsible AI and data ethics.

This dataset is provided “as is” without warranty of any kind. Users assume all risks associated with its use.