Similarity Search in Practice

This bachelor thesis describes the development of a template generation algorithm based on structured product data that are modeled in the organic product catalog Goliath. Goliath was developed during several projects at Otto-von-Guericke University, Magdeburg. The suggested algorithm is an extension for the existing catalog implementation. It will introduce several analysis steps which explore structural as well as semantic aspects and create possibilities to group products based on their semantic similarity. After this, the work defines a format for product description templates and also shows how to extract those by using Greedy algorithms. The templates can be used to simplify product creation for the product creator. Finally, the thesis combines all steps to an overall template generation process and states recommendations for the use of this process in a distributed environment.