Abstract
Implementations of models of morphologically rich languages such as Arabic typically achieve speed and small memory footprint at the cost of abandoning linguistically abstract and elegant representations. We present a solution to modeling rich morphologies that is both fast and based on linguistically rich representations. In our approach, we convert a linguistically complex and abstract implementation of Arabic verbs in finite-state machinery into a simple precompiled tabular representation.
Original language | English (US) |
---|---|
Pages | 116-124 |
Number of pages | 9 |
State | Published - 2011 |
Event | 9th International Workshop Finite State Methods and Natural Language Processing, FSMNLP 2011 - Blois, France Duration: Jul 12 2011 → Jul 15 2011 |
Conference
Conference | 9th International Workshop Finite State Methods and Natural Language Processing, FSMNLP 2011 |
---|---|
Country/Territory | France |
City | Blois |
Period | 7/12/11 → 7/15/11 |
ASJC Scopus subject areas
- Computational Theory and Mathematics
- Language and Linguistics
- Linguistics and Language