Problem and Solution Framing in Senate Tweets: Scaling Expert Interpretation With Machine Learning

ABSTRACT Central to theories of agenda setting is the distinction between problems and solutions. Kingdon's multiple streams framework argues these processes are rhetorical and have distinct logics and incentives, yet empirical research struggles to measure them separately in digital, elite communication. Legislators bridge the multiple streams through their representational styles and legislative behavior, yet we lack scalable tools for identifying when elites frame issues as problems and when they promote solutions. This research note introduces a machine learning approach for classifying elite communication according to this core agenda‐setting distinction. We develop and validate a supervised model for 1.68 million tweets from U.S. Senators that categorizes messages into problem and solution frames, demonstrating that the problem/solution distinction can be reliably automated. This methodological research note establishes that the distinction is empirically measurable at scale and offers a validated, dataset‐specific measurement approach and theoretical foundation for classifying senatorial Twitter communication as problems or solutions.