RAG and LLM Architectures for Question Generation: A Systematic Mapping Study

This dataset contains the complete extraction and analysis data of a Systematic Mapping Study (SMS) on automatic question generation (QG/AQG) with Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG). The search was executed on four databases (Scopus, ACM Digital Library, ScienceDirect, IEEE Xplore), covering November 2022 to 2026. From 947 identified records, 865 remained after deduplication, 175 were selected by title/abstract screening, and 118 studies were included after full-text assessment (identified as s1..s118). The package includes the master extraction table (118 studies x 46 fields), one CSV per research question (SQ1-SQ5), full frequency tables, cross-tabulations and co-occurrence analyses behind the paper figures, and BibTeX entries for all included studies. See README.md for the file dictionary.