I was trying to come up with some good searching examples this morning, and kept running into an old problem with library research – trying to keep authors with similar names separate, while at the same time trying to keep track of researchers who may have published under more than one name. Sure, many resesarchers try to consistently publish under the same name (e.g. continuing to use a maiden name instead of a married name), but sometimes journal formats and database indexers make this difficult: an author may appear as J O'Brien in one journal, JP O Brien in another, and James OBrien in a third. You can find all of these names by using ORs and truncation in your searches (e.g. obrien j* OR o'brien j* OR o brien j* in Web of Science), but then you'll also retrieve everything by John and Joan and Joaquin O'Brien, not to mention that other James O'Brien (I once knew a researcher who'd never had a middle name; he adopted an initial in graduate school in an attempt to differentiate himself from the other researcher in the same field who had the same name). So, what to do? Unfortunately, there are still no good solutions. Many scientists and publishers are advocating the development of a numbering system to keep track of who wrote what – this would be beneficial, but would most likely require each author to ensure that each of their publications is linked to their unique number. Database developers are also working on algorithms to separate papers into groups likely written by the same author, but researchers with name changes, institution changes and new research directions tend to stay a few steps ahead of those algorithms.
If you're interested in some of the potential solutions, take a look at this recent Science article by Martin Enserink (link works on campus only), and http://www.researcherid.com from Thomson-Reuters.
In the meantime, if you are trying to find all the articles by a certain H. Wang or J. Smith, you can try to narrow down your results list by adding other information to your search, such as institutional affiliation, publication years or subject keywords, but be prepared for a mixed bag of results!