Willi Mann successfully defended his PhD Thesis “Efficient Techniques for Set Similarity Join Queries”. While being a student at the Department of Computer Science Willi was associated with the Doctoral College “GIScience” at Z_GIS.
The DK GIScience has profited from this in-depth research in database science. In particular the focus is to efficiently store and query very large volumes of data. The combination of database research and Geoinformatics will provide a deeper understanding of fundamental problems in bridging the gap between database systems and GIS software.
In his Thesis, Willi focused on techniques for set similarity join queries. The join is a fundamental operation in databases. A growing number of problems require joins with fuzzy predicates. A class of frequently applied fuzzy predicates is the similarity of two sets. Typical set similarity functions include Jaccard, Cosine, dice, and Overlap, and the Hamming distance. The set similarity join can be executed in offline settings where the input is known up front and online settings, like on streams of timestamped sets.
Willi’s work resulted in three high level publications which correspond to the three chapters of his thesis: Willi proposed a new filter, PEL, for offline set similarity join algorithms. Then he extensively compared offline set similarity algorithms and finally he proposed a new algorithm, SWOOP, for set similarity joins on streamed data sources.
We wish Willi all the best for his future career and endeavours!