The question for quality in collaborative online projects is a highly discussed topic. Especially in the case of volunteered geographic information (VGI) quality plays a significant role. There exist different quality metrics against which the geographical data can be tested. Examples for these quality aspects are the completeness and correctness of geographic elements themselves and their metadata. Such annotations are called tags and those are entered manually by the volunteers. Different knowledge, level of experience and practice of the contributors are some of the aspects that need to be considered when observing the tag quality. In this thesis, statistics about the users, as well as the tags and the data are presented. Based on these, further investigations regarding quality aspects and improvement are executed. An algorithm is developed that suggests top-k similar sets of tags to the user entering information. In contrast to common recommender system this approach delivers entire tag sets instead of single key-value pairs to the mapper. This empowers the user to select the most comprehensive and adequate tag set. Finally, it is shown that the utilization of the proposed system in mapping routines is a valid method to improve the quality of the tags. This is done with empirical studies and intrinsic comparisons.