96 • 2011 - Hamburger Zentrum für Sprachkorpora - Universität ...

Multilingual Resources and Multilingual Applications - Regular Papers
extensible to more countries. We tested the system with
3 countries with strongly diverging standards for the
expression of location information (Germany, Australia
and Japan). New countries can be added within a few
hours, as only certain country specific files have to be
edited and the corresponding OpenStreetMap knowledge
base has to be plugged in. Most European countries are
similar to Germany, and the U.S. and Canada almost
identical to the Australian system, so that a large part of
the world can easily be covered.
The system was shown to successfully improve the
address element segmentation in a company internal
database with high variation in orthography and
formatting, even containing translated names.
Moreover, the system is able to almost always correctly
guess the country that textual location information can
be attributed to.
In future work, the system can be further improved to
deal with a greater variety of typographical or
transcription errors by using phonetic indexing
algorithms as Soundex for English or Traphoty matching
rules (Lisbach, 2010) for international languages.
6. Acknowledgements
We would like to thank all external annotators that
helped gathering and annotating the test data and the
LSS R&D GmbH for making a company internal
address database available to us in order to test the
system.
7. References
Agichtein, E., Ganti, V. (2004): Mining Reference
Tables for Automatic Text Segmentation. In KDD ’04:
Proceedings of the tenth ACM SIGKDD international
conference on Knowledge discovery and data mining,
Seattle, WA, USA, ACM.
Appelt, D.E., Hobbs, J.R., Bear, J., Israel, D., Tyson, M.
(1992): FASTUS: A Finite-state Processor for
Information Extraction from Real-world Text.
Borkar, V.; Deshmukh, K., Sarawagi, S. (2001):
Automatic segmentation of text into structured
records .
Christen, P., Belacic, D. (2005): Automated Probabilistic
Address Standardisation and Veri- fication.
Australasian Data Mining Conference 2005
(AusDM05).
Christen, P.; Churches, T., Zhu, J.X. (2002): Case-
Probabilistic Name and Address Cleaning and
Standardisation. The Australasian Data Mining
Workshop 2002.
Cortez, E., De Moura, E.S. (2010): ONDUX: On-
Demand Unsupervised Learning for Information
Extraction. In Proceedings of the 2010 international
conference on Management of data (SIGMOD ’10 ),
pp. 807–818.
Lisbach, B. (2010): Linguistisches Identity Matching.
Vieweg+Teubner. ISBN 978-3-8348-9791- 6. URL
http://dx.doi.org/10.1007/ 978-3-8348-9791-6\_11.
Manning, C.D., Schütze, H. (1999): Foundations of
Statistical Natural Language Processing. The MIT
Press, Cambridge, Massachusetts.
Marques, N.C., Gon Calves, S. (2004): Applying a Partof-Speech
Tagger to Postal Address Detection on the
Web, 2004.
Peng, F., McCallum, A. (2003): Accurate In- formation
Extraction from Research Papers using Conditional
Random Fields. In: Information Processing
Management.
Riloff, E. (1993): Automatically Constructing a Dictionary
for Information Extraction Tasks, AAAI
Press / MIT Press. pp. 811–816.
Wolf, J. (2011): Classifying the components of textual
location information. Diploma Thesis, Department für
Linguistik, Universität Potsdam.
43