Kay-Michael Würzner

Text Resources and Lexical Knowledge: Selected Papers from the 9th Conference on Natural Language Processing KONVENS 2008

Research in Natural Language Processing is critically dependent on adequate electronic text resources and a rich representation of lexical knowledge. Their dynamic interaction is explored in this collection of papers, selected from contributions to the 9th conference on Natural Language Processing (KONVENS 2008).

Joining composition and trimming of finite-state transducers

The composition of two (weighted) finite-state transducers is usually carried out in two steps: In the first step, all accessible states of the result are constructed regardless of their co-accessibility. Non-coaccessible states are removed afterwards in the second step. This approach can lead to huge intermediate automata with only a fraction of their states being useful in the end. We present a novel composition algorithm which avoids the construction of non useful states by using a single depth-first traversal while having the same asymptotic complexity as the existing approaches.