Skip to contents

Generate Topic Labels Using AI

Source: R/topic_modeling.R
generate_topic_labels.Rd

This function generates descriptive labels for each topic based on their top terms using AI providers (OpenAI, Gemini, or Ollama).

Usage

generate_topic_labels(
  top_topic_terms,
  provider = "auto",
  model = NULL,
  system = NULL,
  user = NULL,
  temperature = 0.5,
  api_key = NULL,
  openai_api_key = NULL,
  verbose = TRUE
)

Arguments

top_topic_terms

A data frame containing the top terms for each topic.

provider

AI provider to use: "auto" (default), "openai", "gemini", or "ollama". "auto" will try Ollama first, then check for OpenAI/Gemini keys.

model

A character string specifying which model to use. If NULL, uses provider defaults: "gpt-4.1-mini" (OpenAI), "gemini-2.5-flash" (Gemini), or recommended Ollama model.

system

A character string containing the system prompt for the API. If NULL, the function uses the default system prompt.

user

A character string containing the user prompt for the API. If NULL, the function uses the default user prompt.

temperature

A numeric value controlling the randomness of the output (default: 0.5).

api_key

API key for OpenAI or Gemini. If NULL, uses environment variable. Not required for Ollama.

openai_api_key

Deprecated. Use api_key instead. Kept for backward compatibility.

verbose

Logical, if TRUE, prints progress messages.

Value

A data frame containing the top terms for each topic along with their generated labels.

See also

Other topic-modeling: analyze_semantic_evolution(), assess_embedding_stability(), assess_hybrid_stability(), auto_tune_embedding_topics(), calculate_assignment_consistency(), calculate_eval_metrics_internal(), calculate_keyword_stability(), calculate_semantic_drift(), calculate_topic_probability(), calculate_topic_stability(), find_optimal_k(), find_topic_matches(), fit_embedding_model(), fit_hybrid_model(), fit_temporal_model(), get_topic_prevalence(), get_topic_terms(), get_topic_texts(), identify_topic_trends(), plot_model_comparison(), plot_quality_metrics(), run_contrastive_topics_internal(), run_neural_topics_internal(), run_temporal_topics_internal(), validate_semantic_coherence()

Examples

if (FALSE) { # \dontrun{
top_topic_terms <- get_topic_terms(stm_model, top_term_n = 10)

# Auto-detect provider (tries Ollama -> OpenAI -> Gemini)
labels <- generate_topic_labels(top_topic_terms)

# Use specific provider
labels_ollama <- generate_topic_labels(top_topic_terms, provider = "ollama")
labels_openai <- generate_topic_labels(top_topic_terms, provider = "openai")
labels_gemini <- generate_topic_labels(top_topic_terms, provider = "gemini")
} # }