Understanding how political attention is divided and over what subjects is\ncrucial for research on areas such as agenda setting, framing, and political\nrhetoric. Existing methods for measuring attention, such as manual labeling\naccording to established codebooks, are expensive and can be restrictive. We\ndescribe two computational models that automatically distinguish topics in\npoliticians' social media content. Our models---one supervised classifier and\none unsupervised topic model---provide different benefits. The supervised\nclassifier reduces the labor required to classify content according to\npre-determined topic list. However, tweets do more than communicate policy\npositions. Our unsupervised model uncovers both political topics and other\nTwitter uses (e.g., constituent service). These models are effective,\ninexpensive computational tools for political communication and social media\nresearch. We demonstrate their utility and discuss the different analyses they\nafford by applying both models to the tweets posted by members of the 115th\nU.S. Congress.\n