An AI-Assisted Workflow For Supporting Graduate Researchers With Data Discovery

Graduate researchers at my institution have sporadically requested assistance with finding datasets through reference chat, email communications and research consultations. The underlying challenge with these requests is often more complex than locating a data source. Researchers tend to start with overly broad concepts (i.e. health disparities, political corruption, etc.) that do not directly translate into measurable variables, or have overly specific research topics with too many variables that are difficult to find matching datasets for. This gap between research questions and available data can create barriers for researchers who are unfamiliar with appropriate search strategies.

This presentation describes a workflow for incorporating generative AI into dataset discovery. Developed in response to patterns observed in research requests, the workflow helps researchers move from research questions to operationalized concepts, potential variables, and more effective dataset searches. AI is used as a brainstorming and translation tool to help identify possible constructs, terminology, and measures, while librarian expertise remains essential for evaluating dataset documentation, assessing relevance, and determining whether a data source is appropriate for a research question. I will also share an overview of the AI Dataset Discovery webinar I created, which was carried out as part of a the Library’s AI Toolkit Series, and featured a collaboration with an Academic Coach who provided information on how to access and clean data as part of the search workflow.

Learning Objectives:

·         Participants will be made aware of common barriers graduate researchers face when searching for datasets.

·         Participants will be able to use AI to operationalize variables and find datasets.

·         Participants will recognize opportunities for non-data librarians to serve as intermediaries between researchers, datasets, and external organizations.