Authors of scholarly publications state their affiliation in various forms. This kind of heterogeneity makes bibliographic analysis tasks on institutions impossible unless a comprehensive cleaning and consolidation of affiliation data is performed. We investigate automatic approaches to consolidate affiliation data to reduce manual work and support scalability of affiliation analysis. In particular, we propose to set up a reference database of affiliation strings found in publications. A key step in this task is the matching of different affiliation strings to determine whether or not they match. For affiliation matching we investigate web based similarity measures utilizing the cognitive power of current search engines. They determine the similarity of affiliations based on how the URLs in the result sets of affiliation web searches overlap. We evaluate the effectiveness of affiliation matching based on URL overlap as well as for the combined use with the Soft TF-IDF similarity measure.