Package index • mintR

Preparing texts

Clean texts, find their languages and their duplicates, and cut long texts or pool short ones into documents of a useful length.

Count the words and expressions of each document, find the ones that characterise it, filter the vocabulary, and build the input of a topic model.

extract_ngrams(): Extract n-grams from texts and count them by document
score_collocations(): Score candidate collocations of a corpus
select_collocations(): Select collocations with a rule
compare_collocation_rules(): Compare rules for selecting collocations
compute_tf_idf(): Term frequency-inverse document frequency (tf-idf)
filter_group_quantile(): Keep the rows above a quantile of a metric, within each group
filter_vocabulary(): Filter the vocabulary and report what each rule removes
vocabulary_diagnostics(): What each filtering step removes from the vocabulary
plot_vocabulary_diagnostics(): Plot what each filtering step removes
tokens_to_stm(): Build the input of stm() from a long token table

Fit many candidate models (several vocabularies, numbers of topics and random starts) and compute the quality measures proposed in the literature.

evaluate_topic_models(): Evaluate topic models over corpora and numbers of topics
topic_coherence(): Coherence of each topic's top words
topic_exclusivity(): Exclusivity of each topic's top words
topic_diversity(): Topic diversity
topic_similarity(): Similarity between topics
ldatuning_metrics(): Number-of-topics criteria of Arun et al. (2010) and Deveaud et al. (2014)
topic_stability(): Stability of topics across runs
match_topics(): Match the topics of two models
topic_diagnostics(): Diagnostics of each topic of a model
model_metrics(): Summary metrics of one fitted model

Figures to compare candidate models and to inspect single topics.

plot_model_metrics(): Metrics of a topic model evaluation by number of topics
plot_k_frontier(): Coherence-exclusivity frontier of a grid of topic models
plot_topic_diagnostics(): Coherence and exclusivity of each topic of one model
plot_topic_stability(): Stability of topics across runs, by number of topics
pareto_frontier(): Pareto frontier of two criteria
coherence_exclusivity_score(): Weighted harmonic mean of coherence and exclusivity

Build tests in which coders spot the intruder among a topic’s words or topics, give them the sheets, and score their answers.

Describe each topic by its most characteristic words, by labels and by the documents where it dominates.

topic_terms(): Top terms of each topic, by one or several measures
term_counts(): Term counts of an STM corpus
topic_labels(): Short topic labels from an STM model
topic_documents(): Representative documents of each topic
topic_word_mass(): Probability mass of a set of words in each topic
marginal_topic_model(): Collapse a content-covariate STM to one word distribution per topic
topic_summary(): One-table summary of the topics of a model
topic_correlation(): Correlations between topics
topic_prevalence_by(): Average topic proportions by group of documents

Follow the attention given to each topic, for instance over time.

topic_effects(): Simulated topic effects from an STM estimateEffect() object
plot_topic_prevalence(): Average topic prevalence of an STM model
plot_topic_prevalence_over_time(): Topic prevalence over a continuous covariate

mintR mintR-package: mintR: text mining and structural topic model helpers
calculate_beta() calculate_frex() calculate_score() calculate_lift() extract_top_terms() compute_mean_frex() average_frex() compute_stm_term_counts() stm_term_counts() make_topic_labels() top_documents() extract_topic_effects(): Functions removed in the development version of mintR