Skip to contents

RAG-Enhanced Semantic Search

Source: R/semantic_analysis.R
run_rag_search.Rd

Simple in-memory RAG (Retrieval Augmented Generation) for question-answering over document corpus with source attribution. Uses local (Ollama) or cloud (OpenAI/Gemini) embeddings for semantic search and LLM for answer generation.

Usage

run_rag_search(
  query,
  documents,
  provider = c("ollama", "openai", "gemini"),
  api_key = NULL,
  embedding_model = NULL,
  chat_model = NULL,
  top_k = 5
)

Arguments

query

Character string, user question

documents

Character vector, corpus to search

provider

Character string, provider: "ollama" (local), "openai", or "gemini"

api_key

Character string, API key for cloud providers (or from OPENAI_API_KEY/GEMINI_API_KEY env). Not required for Ollama.

embedding_model

Character string, embedding model. Defaults: "nomic-embed-text" (ollama), "text-embedding-3-small" (openai), "gemini-embedding-001" (gemini)

chat_model

Character string, chat model. Defaults: "tinyllama" (ollama), "gpt-4.1-mini" (openai), "gemini-2.5-flash" (gemini)

top_k

Integer, number of documents to retrieve (default: 5)

Value

List with:

  • success: Logical

  • answer: Generated answer

  • confidence: Confidence score (0-1)

  • sources: Vector of source document indices

  • retrieved_docs: Retrieved document chunks

  • scores: Similarity scores

Details

Simple RAG workflow:

  1. Generate embeddings for documents and query

  2. Find top-k similar documents via cosine similarity

  3. Generate answer using LLM with retrieved context

See also

Other ai: call_gemini_chat(), call_llm_api(), call_ollama(), call_openai_chat(), check_ollama(), describe_image(), generate_topic_content(), get_api_embeddings(), get_best_embeddings(), get_content_type_prompt(), get_content_type_user_template(), get_recommended_ollama_model(), list_ollama_models()

Examples

if (FALSE) { # \dontrun{
documents <- c(
  "Assistive technology helps students with disabilities access curriculum.",
  "Universal Design for Learning provides multiple means of engagement.",
  "Response to Intervention uses tiered support systems."
)

# Using local Ollama (free, private)
result <- run_rag_search(
  query = "How does assistive technology support learning?",
  documents = documents,
  provider = "ollama"
)

# Using OpenAI (requires API key)
result <- run_rag_search(
  query = "How does assistive technology support learning?",
  documents = documents,
  provider = "openai"
)

if (result$success) {
  cat("Answer:", result$answer, "\n")
  cat("Sources:", paste(result$sources, collapse = ", "), "\n")
}
} # }