FAQ

What technologies does TurboCurator by ICPSR use?

TurboCurator is a web-based application that offers metadata recommendations for non-sensitive or restricted data leveraging OpenAI’s ChatGPT.

As TurboCurator by ICPSR matures, the system will support more complex logic and processing. This is a rapidly developing landscape with emerging technology. One of our longer term goals include providing keyword suggestions from the data itself.

When can I submit my Dataverse dataset or ICPSR Study to TurboCurator for recommendations?

Dataverse users after completing the minimal required fields to create a Dataverse dataset, can use TurboCurator to generate recommendations. You will be prompted in TurboCurator to provide additional text from other sources that can be used to generate suggestions.

ICPSR depositors will be prompted to improve their study deposit as part of Deposit Manager in the new Single Stream Deposit Manager tool (available in 2025 for select pilot archives.)

Why is TurboCurator using Open-AI’s ChatGPT?

A series of experiments used Open-AI’s ChatGPT to produce recommendations from ICPSR data, with promising results. We are excited to partner with Dataverse to offer this tool to the Dataverse Community and introduce TurboCurator to ICPSR depositors as part of the new Single Stream Deposit Manager tool (available in 2025 for select pilot archives.)

If I am a Dataverse user, do I need to login to use TurboCurator?

No, you are already logged in to your Dataverse installation and we pass your metadata to TurboCurator securely.

How long will my TurboCurator recommendation be available?

TurboCurator recommendations are available for 60 minutes before the session times out. You are welcome to generate a new recommendation at any time.

What browsers support TurboCurator by ICPSR?

TurboCurator is designed to work in all modern browsers commonly used for PCs, Macs, and mobile devices.

For optimal performance, we recommend the following browsers: Chrome, Firefox and Safari.

How are suggestions formed?

TurboCurator by ICPSR uses the following input:

1. Data depositor’s metadata entries (e.g. title, description/summary, and keywords/ICPSR subject terms)
2. Additional text entered in TurboCurator from research sources

Recommendations are generated using OpenAI’s ChatGPT technology and ICPSR’s suggestion logic.

What is TurboCurator using from my metadata?

Studies and datasets that have sensitive metadata should not use the TurboCurator tool.

For Dataverse users, TurboCurator requests metadata from your Dataverse dataset.

For ICPSR users, TurboCurator is embedded into the Deposit Manager for new Single Stream Deposits.

TurboCurator will send the metadata to OpenAI's ChatGPT in order to generate suggestions for: title, description/summary, and keywords/ICPSR Subject Terms.

The TurboCurator system will record the original metadata and the corresponding suggestions, no additional metadata will be saved.

The TurboCurator system will record the original metadata and the corresponding suggestions, no additional metadata will be saved.