Word2Vec

Synopsis

Computes word embeddings using the CONTES/Gensis implementation.

This module is experimental.

Description

Computes word embeddings using the CONTES/Gensis implementation.

Snippet

📋

<word2vec class="Word2Vec">
    <contesDir></contesDir>
    <python3Executable></python3Executable>
    <workers></workers>
</word2vec>

Mandatory parameters

contesDir

Root directory of CONTES.

python3Executable

Path to the Python 3 executable.

workers

Use this many worker threads to train the model (=faster training with multicore machines).

Optional parameters

additionalArguments

UNDOCUMENTED

jsonFile

File where to write embeddings as a JSON object.

modelFile

UNDOCUMENTED

txtFile

File where to write embeddings as a table.

vectorFeature

Name of the feature where to store embeddings of each token. If this parameter is not set, then embeddings are not stored in any feature.

documentFilter

Only process document that satisfy this expression.

formFeature

Feature to use as word form.

minCount

UNDOCUMENTED

sectionFilter

Process only sections that satisfy this expression.

sentenceLayer

Name of the layer containing sentence annotations.

tokenLayer

Name of the layer containing token annotations.

vectorSize

The dimensionality of the feature vectors. Often effective between 100 and 300.

windowSize

The maximum distance between the current and predicted word within a sentence.

Deprecated parameters

formFeatureName

Deprecated alias for formFeature .

sentenceLayerName

Deprecated alias for sentenceLayer .

tokenLayerName

Deprecated alias for tokenLayer .

vectorFeatureName

Deprecated alias for vectorFeature .