Assistant Professor Ruodan Zhang Uses Machine Learning to Improve the Retrievability of IRS 990 Text Data

Ruodan Zhang
Assistant Professor Ruodan Zhang

Department of Public Policy (DPP) Assistant Professor Ruodan Zhang is working with Haohan Chen, a postdoctoral fellow at the Center for Social Media and Politics, New York University, to develop a method that helps researchers and practitioners identify nonprofit organizations of interest. This research project takes advantage of the newly available nonprofit mission and program service activity description data reported in the IRS Form 990s, and applies word embedding methods to generate a continuous measure for document-to-keyword relevance. DPP graduate student Jonathan Richter (MPA ’22) manually validated the results generated for identity-related search terms. As a next step, Ruodan and Haohan are working on developing a web-based Shiny app for nonprofit researchers and practitioners to further explore and test the identification method.

In 2019 Ruodan and Haohan presented “Identifying Nonprofits by Scaling Mission and Activity with Word Embedding” at the 2019 Association of Public Policy Analysis Management Annual Conference. They published the paper again in November of 2020 after feedback from the panel and Dr. Yuan Cheng. DPP Graduate Assistant Jonathan Richter also provided research assistance for this updated paper.