Partitioning Multitape Transducers
- 510 Downloads
In this paper, we define a class of transducers closed under intersection and complementation, which are the operations used for contextual rule compilation. This class of transducers is not theoretically more powerful than the Epsilon-Free Letter Transducers most commonly used. But they are more convenient for morphological description whenever the correspondence between lexical and surface forms is not a symbol-to-symbol matching.
A complete set of operations on transducers is defined, including some operations (projection and join) which change the number of tapes of the transducers.
Unable to display preview. Download preview PDF.
- 1.Ganchev, H., Mihov, S., Schulz, K.U.: One-letter automata: How to reduce k tapes to one. Technical Report IS-Bericht-03-133, Centrum für Informations- und Sprachverarbeitung, Universität München, Munich (Germany) (2003)Google Scholar
- 2.Grimley-Evans, E., Kiraz, G., Pulman, S.: Compiling a partition-based two-level formalism. In: COLING, Copenhagen, Denmark, pp. 454–459 (1996)Google Scholar
- 3.Kaplan, R.M., Kay, M.: Regular models of phonological rule systems. Computational Linguistics 20(3), 331–378 (1994)Google Scholar
- 4.Kempe, A., Guingne, F., Nicart, F.: Algorithms for weighted multi-tape automata. Technical Report 2004/031, XRCE, Grenoble, France (2004)Google Scholar