Personal names are fundamental in our civilization. Names serve to refer to individuals, but in contrast to e.g. social assurance numbers that are unique for each citizen, names are not treated that strictly. Names do not identify persons in a non-ambiguous way. There are people that share the same name. Furthermore, the fact that names are not treated as carefully as numbers often leads to misspellings and confusion. The increasing importance of the Web in our lives has as consequence that people are more frequently confronted with names of things, places and persons. On the one hand, the Web provides access to more information sources and by this to more information, on the other hand the search for relevant information is getting more difficult. A particular problem occurs in (digital) libraries. They are expected to catalog publications in a convenient way that facilitates and supports literature research. It is necessary to distinguish authors that are referred to by their names. The problem of homonymous authors arises, i.e. there may be several authors sharing a name. Inversely, authors may publish under different names or name variations, deliberately or unintendedly. Digital libraries spend much effort on the disambiguation of author names. This work reports the results of a literature research focusing on disambiguation of homonymous authors and proposes a different perception of the data that is processed during name disambiguation. Different pieces of information are integrated into a graph based approach. An implementation of some ideas of the presented approach serves as a proof of concept and gives further insights in the nature of name disambiguation.