The author name disambiguation process compares citations with the same author name. The similarity for each citation pair is measured by examining the metadata for both citations, such as co-authors, journal, title, affiliation, abstract, MeSH terms, grants, and publication date. Citations that share like author names are divided into different groups by clustering the citations that are highly similar to each other. Citations within each group are then classified as belonging to the same author.

This should be interesting to try out… author disambiguation is a tricky business.

KraftyLibrarian raised a good point:

Since most people tend to publish in their own area of expertise this should be helpful.  The only thing you have to do is make sure you account for those times where they publish outside of their usual field. 

(Source: twitter.com)