Enterprise collaboration platforms are large-scale information infrastructures that provide a wide range of tools and functionality to support collaborative work in organizations. These collaborative activities leave digital traces in the form of social documents, which can be analyzed to understand how employees work together to coordinate their joint work. In this paper, we present the findings of a research project to visualize the structure of social documents to prepare them for analysis as traces of collaborative activity. Using the representation of social documents defined in the Social Document Ontology (SocDOnt), we draw on concepts from graph theory to develop a method for the graphical visualization of social documents. Applying this method to analyze the social documents in an operational enterprise collaboration platform, we identify and display different types of social documents and define their characteristic structure. Our findings provide the necessary foundation for conducting computational ethnographies of collaborative work.Social documents as digital traces of collaborative activity are complex, compound documents created by the use of social software and composed of heterogeneous components [11]. For the investigation of their structure, Hausmann and Williams [11] derived a conceptual information model and provided a first structural description of their nature and a list of possible components, such as versions, comments, attachments, tags