---
name: "mbtg"
section: "1"
platform: "linux"
package: "mbt"
version: "3.11-2"
release: "Debian 13 (Trixie)"
platforms: ["linux"]
url: "https://new.carta.tech/mbtg"
---

# mbtg(1)

Memory Based Tagger generator

## NAME

MBTG - Memory Based Tagger generator

## SYNOPSIS

mbtg -T \<filename\> -s \<setting filename\>

or

mbtg \[options\]

## DESCRIPTION

This programs generates, based on a tagged corpus, all the files needed to be able to tag a text with **mbt.**

## OPTIONS

- **-h** or **--help**

  show help

**-T** \<tagged training corpus file\>

or

**-E** \<enriched tagged training corpus file\>

All further options have reasonable defaults, so using them is only needed for the experienced user. See the mbt manual for more details.

- **-s** settingsfile

  **mbtg** creates this file, which can be used to run **mbt** with minimal effort. (like mbt -s settings -T somefile)

- **-p** pattern

  the pattern for known words (default ddfa)

- **-P** pattern

  the pattern for unknown words (default dFapsss)

- **-%** \<number\>

  filter threshold for ambitag construction (default 5%)

**-l** \<lexiconfile\>

**-L** \<file with list of frequent words\>

**-r** \<ambitagfile\>

**-k** \<known words case base\>

**-u** \<unknown words case base\>

**-K** \<known words instances file\>

**-U** \<unknown words instances file\>

- **-V** or **--version**

  show version info

**-e** \<sentence delimiter\> (default '\<utt\>')

- **-X**

  keep the intermediate files

- **-O**timbl options

   (Note: there is NO SPACE between O and the options)\
     \<options\>   classifier options for both known and unknown words instances bases\
     K: \<options\>   classifier options for known words instance base\
     U: \<options\>   classifier options for unknown words case base\
     valid **timbl** options are: a d k m q v w x -

## BUGS

possibly

## AUTHORS

Ko van der Sloot Timbl@uvt.nl

Antal van den Bosch Timbl@uvt.nl

## SEE ALSO

[**timbl**(1)](/timbl/1.md) [**mbt**(1)](/mbt/1.md) [**mbtserver**(1)](/mbtserver/1.md)