Frequent successful publications by specific institutions are indicators for identifying outstanding centres of research. This institution data are present in scholarly papers as the authors‟ affilations – often in very heterogeneous variants for the same institution across publications. Thus, matching is needed to identify the denoted real world institutions and locations. We introduce an approximate string metric that handles acronyms and abbreviations. Our URL overlap similarity measure is based on comparing the result sets of web searches. Evaluations on affiliation strings of a conference prove better results than soft tf/idf, trigram, and levenshtein. Incorporating the aligned affiliations we present top institutions and countries for the last 10 years of SIGMOD.