Function reference

All functions

as_embed() as_wordvec() `[`(<embed>) pattern()

Word vectors data class: wordvec and embed.

cosine_similarity() cos_sim() cos_dist()

Cosine similarity/distance between two vectors.

data_transform()

Transform plain text of word vectors into wordvec (data.table) or embed (matrix), saved in a compressed ".RData" file.

data_wordvec_load() load_wordvec() load_embed()

Load word vectors data (wordvec or embed) from ".RData" file.

data_wordvec_subset() subset(<wordvec>) subset(<embed>)

Extract a subset of word vectors data (with S3 methods).

demodata

Demo data (pre-trained using word2vec on Google News; 8000 vocab, 300 dims).

dict_expand()

Expand a dictionary from the most similar words.

dict_reliability()

Reliability analysis and PCA of a dictionary.

get_wordvec()

Extract word vector(s).

most_similar()

Find the Top-N most similar words.

normalize()

Normalize all word vectors to the unit length 1.

orth_procrustes()

Orthogonal Procrustes rotation for matrix alignment.

pair_similarity()

Compute a matrix of cosine similarity/distance of word pairs.

plot_network()

Visualize a (partial correlation) network graph of words.

plot_similarity()

Visualize cosine similarity of word pairs.

plot_wordvec()

Visualize word vectors.

plot_wordvec_tSNE()

Visualize word vectors with dimensionality reduced using t-SNE.

sum_wordvec()

Calculate the sum vector of multiple words.

tab_similarity()

Tabulate cosine similarity/distance of word pairs.

test_RND()

Relative Norm Distance (RND) analysis.

test_WEAT()

Word Embedding Association Test (WEAT) and Single-Category WEAT.

text_init()

Install required Python modules in a new conda environment and initialize the environment, necessary for all text_* functions designed for contextualized word embeddings.

text_model_download()

Download pre-trained language models from HuggingFace.

text_model_remove()

Remove downloaded models from the local .cache folder.

text_to_vec()

Extract contextualized word embeddings from transformers (pre-trained language models).

text_unmask()

<Deprecated> Fill in the blank mask(s) in a query (sentence).

tokenize()

Tokenize raw text for training word embeddings.

train_wordvec()

Train static word embeddings using the Word2Vec, GloVe, or FastText algorithm.