;;; -*- Mode: tdl; Coding: utf-8; indent-tabs-mode: nil; -*-

;;; Inbuilt tagging in PET is controlled firstly by the `taggers' setting.
;;; This controls which taggers are run in which order. The tagger names 
;;; are arbitrary strings which are used as a prefix for the tagger settings
;;; that control how the taggers are run, what input they require, and how the
;;; output is processed. Relevant settings are listed below:
;;;  * <prefix>-command: REQ, command line for tagger
;;;  * <prefix>-tok_sep: REQ, string that separates tokens
;;;  * <prefix>-utt_sep: REQ, string that separates items
;;;  * <prefix>-tag_format: REQ, specifies the output of the tagger, one of:
;;;		* "single" - one tag per token, format: word tag
;;;      * "multi" - multiple tags per token, format: word (tag prob)+
;;;      * "genia" - genia output, format: word lemma POS chunk NE
;;;  * <prefix>-arguments: OPT, command line arguments for tagger
;;;  * <prefix>-mapping: OPT, words that should be mapped before input to tagger
;;;  * <prefix>-utterance-start: OPT, token that should be given to tagger
;;;    before each item (for taggers that assume continuous input)
;;;  * <prefix>-utterance-end: OPT, token that should be given to tagger
;;;    after each item (for taggers that assume continuous input)
;;;  * <prefix>-pos_sep: OPT, string to separate token and POS in tagger input.
;;;    Non-empty pos_sep indicates that the tagger expects token and POS as
;;;    input.
;;;  * <prefix>-namedentities: OPT, controls whether named entities are added as
;;;    extra tokens. Ignored if tag_format != "genia".

taggers := tnt.
;taggers := tnt stnt.
tnt-command := "${LOGONROOT}/bin/tnt".
tnt-arguments := "-v0 -z100 ${LOGONROOT}/coli/tnt/models/wsj -".
tnt-tok_sep := "
".
tnt-utt_sep := "

".
tnt-tag_format := "multi".
tnt-utterance-start := ".".
tnt-utterance-end := ".".
tnt-mapping := 
  "“" "``" 
  "”" "''" 
  "‘" "`" 
  "’" "'"
  "…" "..."
  "—" "---"
  "–" "--".
stnt-command := "${LOGONROOT}/bin/tnt".
;stnt-arguments := "-v0 -z100 ${HOME}/work/supertagging/mapped_models/not-wsj02_model -".
stnt-arguments := "-v0 -z100 ${HOME}/work/supertagging/morph_models/not-wsj02_morphmodel -".
stnt-tok_sep := "
".
stnt-utt_sep := "

".
stnt-tag_format := "multi".
stnt-utterance-start := ".".
stnt-utterance-end := ".".
stnt-mapping := 
  "“" "``" 
  "”" "''" 
  "‘" "`" 
  "’" "'"
  "…" "..."
  "—" "---"
  "–" "--".
genia-command := "${LOGONROOT}/bin/geniatagger".
genia-arguments := "-nt".
genia-tok_sep := " ".
genia-utt_sep := "
".
genia-tag_format := "genia".
genia-mapping := 
  "“" "``" 
  "”" "''" 
  "‘" "`" 
  "’" "'"
  "…" "..."
  "—" "---"
  "–" "--".
genia-namedentities := no.