The list cleanup tool is actually quite nice and for affiliation data you have two fuzzy algorithms you can use (dept level or higher). Of course cleaning is time-consuming; it always is. And, an algorithm can only do so much so there is always an amount of human judgement necessary. The best approach is to check what level you want and let the algo do its work. Next you display all items and alphabetize these. Check all names carefully so you get an idea of what is there. If you see any organizations which are duplicates you can cut and paste these into the affiliation name you would like to collect all iterations of the organization under. Alternatively you can drag and drop. Once satisfied with the level of cleaning save the clean-up as a thesaurus for future use and accept (!). This will make your clean-up a new field (as can be seen in the data-set summary). Good luck!

Just a simple remark - cleaning data takes even 80% time of bibliometric study. In light of the above I think that patience might be the best word for this level. Currently, I am before data cleaning in my research. However, I am reading papers about data cleaning. If you are interested I can send you references, if you have your own list of papers about data cleaning I would be interested in it. All the best, M.