Statistical and Machine Learning Approaches for Network Analysis

Acknowledgments

We thank all members of the department of the Goethe University that contributed to this work, especially Prof. Alexander Mehler. Furthermore, I want to thank Vincent Esche, Dr. Ingo Glöckner, and Armin Hoehnen for proof-reading this document. Also, I am indebted to Prof. Hermann Helbig and Dr. Sven Hartrumpf for letting me use the WOCADI parse of the Wikipedia.

Notes

1. Note that our approach is also used to extract instance-of relations.

2. WOCADI is the abbreviation of word-class disambiguating parser.

3. HaGenLex is the abbreviation of Hagen GermanLexicon.

4. Note that in our evaluation, the index i was started at one instead of zero in order to exploit the entire value range for the normalized kernel value from zero to one.

5. GermaNet synsets were mapped to HaGenLex concepts.