Revision history for Perl extension Alvis::NLPPlatform.
0.3 - additional options for the link parser wrapper, to write link
parser postscritp output (PARSING_IN_P0STSCRIPT)
and/or link parser graphics output (PARSING_GRAPHICS) in file
- bug fix in the default term wrapper (a term embedded in a
named entity was not detected)
- bug fix in the default syntactic parser. Take into account
empty sentence parsing.
- Modification of the TermTagger : term list is loaded once.
- Display of the processing time for each step
- fix a bug in the XML loader of semantic unit/named-entity
- Definition of a section to manage XML input : the option
PRESERVEWHITESPACE is set in; addition of the option
LINGUISTIC_ANNOTATION_LOADING
- Definition of a section to manage XML output
- Render time is saved in the xml file (Client/server and
stantalone mode)
0.2
- rename client.pl, server.pl, go.pl, as alvis-nlp-client, alvis-nlp-server, alvis-nlp-standalone
- set the default parsing wrapper (forgot in the version 0.1)
- configuration files have been moved in the alvis-nlpplatform
- writing functions printing header and footer of the
documentCollection
- managing annoation of several document in the standalone
mode
- correction in the term wrapper, in the generation of the
semantic units
- correction in the default syntactic wrapper, in the
generationo of the head and modifier element. Corrections
also reported in the syntactic wrapper in the UserNLPWrapper
module
- default term tagger wrapper takes into account given
canonical form of terms if the term tagger returns it.
- modified dependencies for syntactic parsing (now needs PoS tagging)
- control characters are not visibles
- addition of an option for preserving or not the white
spaces (XML blank nodes, but also beginning and end of the
line spaces). Default behaviour is removing the white spaces.
- addition of a patch for Link Gramar Parser 4.1a and
4.1b. This patch adds a option (!whitespaces) to avoid LG
tokenization.
0.1
- version with CPAN configuration