Implementing a Chinese Character Browser Using a Topography-Preserving Map
The Chinese Character Browser is a user interface designed for the search and exploration of a database of Chinese characters, Chinese pronunciations (pinyin), and English definitions. The browser uses a technology based upon Kohonen’s self-organizing map to map the 10-dimensional feature vector describing each Chinese character onto a discrete two-dimensional grid, which forms the foundation for the browser. The Chinese Character Browser was designed to demonstrate the importance not only of topology preservation, but topography preservation in such mappings. In brief, to the extent that a mapping is topography-preserving, the structure of the output map grid can reflect the structure of the original data at several levels of granularity simultaneously, allowing the assumption of a hierarchical organization in the output map that corresponds to the structure of the input data set. This can significantly speed up search as well as make the mapping more useful for a variety of applications.
KeywordsChinese Character Input Method Dist Versus Prototype Vector Topology Preservation
Unable to display preview. Download preview PDF.
- 1.Breen, J.: KANJIDIC Documentation (November 2005) . Unpublished, Available at, accessed, http://www.csse.monash.edu.au/~jwb/kanjidic.html
- 2.Halpern, J.: NTC’s New Japanese-English Character Dictionary. National Textbook Co., Lincolnwood, Lincolnwood, Illinois (1993)Google Scholar
- 3.Kirk, J.S.: Algorithms to Improve Topography Preservation in Self-Organizing Maps. Ph.D. dissertation, University of Louisville (2003)Google Scholar
- 4.Kirk, J.S., Zurada, J.M.: An Evolutionary Method of Training Topography-Preserving Maps. In: Proceedings of the 2001 INNS-IEEE International Joint Conference on Neural Networks, Washington, D.C, July 14-19, pp. 2230–2234 (2001)Google Scholar
- 5.Kirk, J.S., Zurada, J.M.: Motivation for a Genetically-Trained Topography-Preserving Map. In: Proceedings of the 2002 International Joint Conference on Neural Networks, Honolulu, HI, May 12-17, pp. 394–399 (2002)Google Scholar
- 9.Koskela, M., Laaksonen, J., Laakso, S., Oja, E.: The PicSOM Retrieval System: Description and Evaluations. In: Eakins, J.P., Enser, P.G.B. (eds.) Proceedings of Challenge of Image Retrieval, Brighton, UK (2000)Google Scholar