Training

Lindera Python supports training custom CRF-based morphological analysis models from annotated corpora. This functionality requires the train feature.

Prerequisites

Build lindera-python with the train feature enabled (enabled by default):

maturin develop --features train

Training a Model

Use lindera.train() to train a CRF model from a seed lexicon and annotated corpus:

import lindera

lindera.train(
    seed="resources/training/seed.csv",
    corpus="resources/training/corpus.txt",
    char_def="resources/training/char.def",
    unk_def="resources/training/unk.def",
    feature_def="resources/training/feature.def",
    rewrite_def="resources/training/rewrite.def",
    output="/tmp/model.dat",
    lambda_=1.0,
    max_iter=100,
    max_threads=4,
)

Training Parameters

Parameter	Type	Default	Description
`seed`	`str`	required	Path to the seed lexicon file (CSV format)
`corpus`	`str`	required	Path to the annotated training corpus
`char_def`	`str`	required	Path to the character definition file (char.def)
`unk_def`	`str`	required	Path to the unknown word definition file (unk.def)
`feature_def`	`str`	required	Path to the feature definition file (feature.def)
`rewrite_def`	`str`	required	Path to the rewrite rule definition file (rewrite.def)
`output`	`str`	required	Output path for the trained model file
`lambda_`	`float`	`1.0`	Regularization coefficient, finite and >= 0; `0` disables regularization for every type; with L2, MeCab's `-c C` is `1 / C`, so 1.0 is MeCab's `-c 1.0`
`max_iter`	`int`	`100`	Maximum number of training iterations, 1 or more; training stops earlier once the objective converges (see `eta`)
`max_threads`	`int` or `None`	`None`	Number of threads, 1 or more (None = auto-detect CPU cores)
`regularization`	`str`	`"l2"`	Regularization type: `"l1"`, `"l2"` or `"elasticnet"`, in any case (`"elastic_net"` and `"elastic-net"` also select Elastic Net), as `lindera train --regularization` takes it
`elastic_net_l1_ratio`	`float`	`0.5`	L1 ratio of Elastic Net regularization, between 0 and 1; `0` uses only L2, `1` uses only L1; used only with `"elasticnet"`
`eta`	`float`	`5e-5`	Tolerance of the convergence test, finite and >= 0: training stops before `max_iter` once the relative change of the objective stays below it for 3 iterations in a row; `0` disables the test

The settings are checked before any file is read: an unknown regularization, or a lambda_, elastic_net_l1_ratio or eta outside its range, raises a ValueError. To train with v6's default settings (the model still differs from v6's, since the first line search changed in #1290), pass regularization="l1" and lambda_=0.01, and eta=0 to stop only at max_iter.

Exporting a Trained Model

After training, export the model to dictionary source files using lindera.export():

import lindera

lindera.export(
    model="/tmp/model.dat",
    output="/tmp/dictionary_source",
    metadata="resources/training/metadata.json",
)

Export Parameters

Parameter	Type	Default	Description
`model`	`str`	required	Path to the trained model file (.dat)
`output`	`str`	required	Output directory for dictionary source files
`metadata`	`str` or `None`	`None`	Path to a base metadata.json file

The export creates the following files in the output directory, the same files that lindera export creates:

lex.csv -- Lexicon entries with trained costs
matrix.def -- Connection cost matrix
unk.def -- Unknown word definitions
char.def -- Character category definitions (copied from the trained model)
feature.def -- Feature template definitions (copied from the trained model)
rewrite.def -- Feature rewrite rules (copied from the trained model)
left-id.def -- Left context ID to feature string mapping
right-id.def -- Right context ID to feature string mapping
metadata.json -- Updated metadata (when metadata parameter is provided)

The model file must come from this version's training. A model file written by an earlier version, which includes every model file written before 7.0.0, is rejected: lindera.export() fails with an error that asks you to re-run training. Train such a model again with lindera.train().

Complete Workflow

The full workflow for training and using a custom dictionary:

import lindera

# Step 1: Train the CRF model
lindera.train(
    seed="resources/training/seed.csv",
    corpus="resources/training/corpus.txt",
    char_def="resources/training/char.def",
    unk_def="resources/training/unk.def",
    feature_def="resources/training/feature.def",
    rewrite_def="resources/training/rewrite.def",
    output="/tmp/model.dat",
    lambda_=1.0,
    max_iter=100,
)

# Step 2: Export to dictionary source files
lindera.export(
    model="/tmp/model.dat",
    output="/tmp/dictionary_source",
    metadata="resources/training/metadata.json",
)

# Step 3: Build the dictionary from exported source files
metadata = lindera.Metadata.from_json_file("/tmp/dictionary_source/metadata.json")
lindera.build_dictionary("/tmp/dictionary_source", "/tmp/dictionary", metadata)

# Step 4: Use the trained dictionary
tokenizer = (
    lindera.TokenizerBuilder()
    .set_dictionary("/tmp/dictionary")
    .set_mode(lindera.Mode.normal())
    .build()
)

tokens = tokenizer.tokenize("形態素解析のテスト")
for token in tokens:
    print(f"{token.surface}\t{','.join(token.details)}")

Lindera Documentation