I would combine Sklivvz suggestion with at least title tag inspection and if possible Whois records.
If you parse the copyright notice of stackoverflow.com you either end up with "stackoverflow.com" or "stackoverflow.com LLC" and that is kind of irrelevant.
In the maxmind.com example if you parse the legal notice "Copyright © 2009 MaxMind, Inc." you either end up with "MaxMind" or "MaxMind, Inc.", if you combine with the title tag "Geolocation and Online Fraud Prevention from MaxMind" the only word that appears at both places is MaxMind.
Also domaintools.com breaks the domain name into words (everyone can change it though), you can parse this page and grab the "real" name, some examples:
http://whois.domaintools.com/maxmind.com = Max Mind
http://whois.domaintools.com/dabbledb.com = Dabble Db
http://whois.domaintools.com/domaintools.com = Domain Tools
http://whois.domaintools.com/stackoverflow.com = Stack Overflow
http://whois.domaintools.com/joelonsoftware.com = Joel On Software
aboutus.org might also be a very good reference when avaliable:
http://aboutus.org/maxmind.com = MaxMind LLC
http://aboutus.org/dabbledb.com = ...
http://aboutus.org/domaintools.com = DomainTools.com
http://aboutus.org/stackoverflow.com = Stack Overflow AG
http://aboutus.org/joelonsoftware.com = Fog Creek Software Inc.