This function creates and trains a BERTopic model (based on bertopic python packaged) on a text-variable in a tibble/data.frame. (EXPERIMENTAL)

textTopics(
  data,
  variable_name,
  embedding_model = "distilroberta",
  umap_model = "default",
  hdbscan_model = "default",
  vectorizer_model = "default",
  representation_model = "mmr",
  num_top_words = 10,
  n_gram_range = c(1, 3),
  stopwords = "english",
  min_df = 5,
  bm25_weighting = FALSE,
  reduce_frequent_words = TRUE,
  set_seed = 8,
  save_dir = "./results"
)

Arguments

data: (tibble/data.frame) A tibble with a text-variable to be analysed, and optional numeric/categorical variables that you might want to use for later analyses testing the significance of topics in relation to these variables.
variable_name: (string) Name of the text-variable in the data tibble that you want to perform topic modeling on.
embedding_model: (string) Name of the embedding model to use such as "miniLM", "mpnet", "multi-mpnet", "distilroberta".
umap_model: (string) The dimension reduction algorithm, currently only "default" is supported.
hdbscan_model: (string) The clustering algorithm to use, currently only "default" is supported.
vectorizer_model: (string) Name of the vectorizer model, currently only "default" is supported.
representation_model: (string) Name of the representation model used for topics, including "keybert" or "mmr".
num_top_words: (integer) Determine the number of top words presented for each topic.
n_gram_range: (vector) Two-dimensional vector indicating the ngram range used for the vectorizer model.
stopwords: (string) Name of the stopword dictionary to use.
min_df: (integer) The minimum document frequency of terms.
bm25_weighting: (boolean) Determine whether bm25_weighting is used for ClassTfidfTransformer.
reduce_frequent_words: (boolean) Determine whether frequent words are reduced by ClassTfidfTransformer.
set_seed: (integer) The random seed for initialization of the umap model.
save_dir: (string) The directory for saving results.

Value

A folder containing the model, data, folder with terms and values for each topic, and the document-topic matrix. Moreover the model itself is returned formatted as a data.frame together with metdata. See textTopicsReduce

textTopicsTest and textTopicsWordcloud.