-
Modified: Wed Feb 8 12:13:36 2006 21:13:35 Uhr MEST (mike)
- the sentence number
- the node number (lexical nodes are counted beginning with 1)
- coreference information of the following kind
- %R1=<key> : first occurrence in a coreference chain
- %R=<key> : further occurrences in a coreference chain
- %Db=<key> : definite description dependent on some other entity (bridging)
- %GENERIC : existential definite
FSA-tools for compilation of large grammars (gzipped tarfile), ask me for the source code if you need it; this program generates a transition table with 1,200,000 entries for a minimal deterministic finite-state automaton in less than 15 minutes on a Sun Blade 1000.
An excerpt of manually and automatically annotated coreference links of the Negra treebank (as described in the 2004 COLING paper). Contact me to get the full body of data. Format: there are three columns:
A perl script and a data file to recognize Chinese numbers (in UTF-8 encoding) and convert them into the Western / Arabic system.