Back to Working Groups

Social Research with Unstructured Data

MIDAS Working Group


Social science increasingly uses data that requires advanced tools and algorithms. Preparing text, audio, and video data for analysis and running those analyses involves linking contemporary computational tools with datasets and major research challenges in ways that cut across disciplines. Building bridges between data scientists who develop the tools to analyze these data and social scientists who have datasets to address research questions in unprecedented ways that could benefit from them is critical for both opening the world of these new data and refining the tools for analysis.

The Social Research with Unstructured Data working group facilitates collaborations between NLP methodology experts and domain researchers who pose significant research questions that could benefit from NLP. One of the primary working group activities is organizing a regular connection series, with presentations on works in progress, funding opportunities, and new data sources. This ongoing series, jointly sponsored by MIDAS, the AI Lab and ISR, began in Fall 2022.

The working group also organized an NLP workshop series in Fall 2022. Past workshop series information is available here.

Working Group Members

Danai Koutra – Associate Director, MIDAS | Associate Professor, Computer Science and Engineering

Josh Pasek – Associate Director, MIDAS | Associate Professor, Communications and Media

Elyse Thulin – Michigan Data Science Fellow

Julia Lippman – Senior Research Specialist, Center for Political Studies

Beth Uberseder – Research Manager, MIDAS

Social Research with Unstructured Data: Connection Series

The Social Research with Unstructured Data Connection Series is designed to build connections between social scientists and data scientists to improve scholarship in both arenas. The series connects faculty, research scientists, postdoctoral, and graduate students to help build major research projects from their vision, find the right methods for their data, identify collaborators, or find scholars who can help trial run the tools they have developed.

For more information, or if you are interested in presenting work in progress at this series, please fill out this form, or contact the MIDAS Research Manager, Beth Uberseder ( For the schedule of upcoming events as well as past speakers for this series, see below.


Upcoming Events

During the Winter 2023 Term, the Social Research with Unstructured Data: Connection Series events are held from 2:00-3:30 PM in ISR 6050 (426 Thompson Street).

January 20, 2023
No presenters this week. Please come prepared with an elevator pitch about your research and join us for a networking and brainstorming event to kick off the series this term.
February 3, 2023
Presenting this week: Walter Mebane discussing “A Twitter Election Observatory Using Decahose Data,” and Viktoryia Kalesnikava and Aparna Ananthasubramaniam presenting on NLP methods focused on understanding circumstances around suicide deaths that occurred in 2020.
March 10, 2023
Presenting this week: Sara Lafia, David Bleckley & Trent Alexander on “Digitizing and parsing semi-structured documents from the GI Bill mortgage program,” as well as Stella Yu (Talk TBA) and David Jurgens (Talk TBA)
March 31, 2023
Presenting this week: Z. Tuba Suzer-Gurtekin, on “Explaining Consumer Expectations using Big Data,” and Dallas Card (Talk TBA)

Past Events and Speakers

November 11, 2022
Fred Conrad, Professor of Psychology and Director of the Survey Research Center, discussing recent work using stance detection to uncover hidden alignment between social media posts and survey responses.
Vinod Vydiswaran, Associate Professor of Learning Health Sciences and the School of Information, presenting “NLP works hand-in-hand with Qualitative Research — Solutions and Pitfalls”.
December 9, 2022
Elle O’Brien, Lecturer & Research Investigator at the School of Information, discussing “Barriers to Adopting Data Science Methods for Research”.
Brady West, Research Professor in the Survey Research Center, on “Adjusting Estimated Regression Coefficients for Mismatch Error when Linking Twitter and Survey Data”.