Here's an example VIAF author cluster:

http://viaf.org/viaf/49466902/viaf.xml

Without getting too far into the details of what data we're looking for or what counts as success, let's start by parsing this into an intermediate format so we can reason about it with software.

Here are the bits that seem most important.

• We have a list of <viaf:data> tags representing different names this person has been given: "Moers, Walter, 1957-....", "Moers, Walter", "Walter Moers deutscher Comic-Zeichner, Illustrator und Autor", and so on. It's possible that these tags just reproduce data from the <viaf:mainHeadingEl> tags later on, in which case we could ignore them.
• Each name has a list of sources associated with it. These are the libraries (generally national libraries) that have assigned this name to the author. We care about this because in most cases we want the names assigned by English-speaking libraries. We don't want to consider "Моэрс, В. 1957- Вальтер" (from the Russian national library), even though that is accurate, because that's not the name most of our patrons would use. (However, if a non-English-speaking library uses the same name as an English-speaking library, that counts as evidence for the quality of the name.)
• We have a list of <viaf:mainHeadingEl> tags representing (I think) various preferred forms of the person's name. These contain name data in an XML MARC format, either regular MARC or UNIMARC.
• We have a list of <viaf:x400> tags representing various 4xx-series MARC records found for this person. These also contain name data in an XML MARC format. These also have source information.
• We have a list of <viaf:x500> tags representing 5xx-series MARC records found for this person. This includes pseudonyms. These tags also come with source information.
• We have a <viaf:coauthors> tag and a <viaf:publishers> which we can skip.
• There's a <viaf:birthDate> and <viaf:deathDate> which might be interesting.
• There are <viaf:xLinks> to Wikipedia pages, which might be useful.