Summarize Paper logo SummarizePaper.com
Premium+

Literature review

Ask a research question. We look for the papers that answer it, find the founding and the recent ones, group them by approach and write a comparative synthesis that cites them.

How to ask your question and check the synthesis? Illustrated guide

A real example: what you get

Example in Astrophysics Computer science and AI Biology and medicine

30 papers in 6 themes · “Retrieval-Augmented Generation and Hallucination Reduction”.

  • A synthesis, each sentence with its papers
  • 6 themes with their findings
  • 3 points of consensus
  • 4 open questions
  • Each statement with the quoted sentence of its paper
The synthesis

The clearest direct comparisons favor retrieval-augmented generation (RAG) over fine-tuning alone for factual question answering, especially for new or less-popular knowledge, but these studies establish answer-performance advantages rather than a uniform, directly measured reduction in hallucination.

A point of consensus, with its source

The direct comparative evidence favors RAG over fine-tuning as a method for accessing factual knowledge, particularly when facts are new or uncommon.

Its source Notably, when comparing knowledge injection methods, RAG significantly outperforms FT. — Fine Tuning vs. Retrieval Augmented Generation for Less Popular Knowledge

The first results

  1. Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni et al. · 2020 It establishes RAG as a knowledge-access approach and evaluates knowledge-intensive tasks, but the abstract does not make a direct fine-tuning-only comparison.
  2. Retrieval Augmentation Reduces Hallucination in Conversation Kurt Shuster, Spencer Poff, Moya Chen, Douwe Kiela et al. · 2021 Shows retrieval can improve knowledge-grounded dialogue, but does not compare against fine-tuning.

28 more results hidden

See a complete example, explained step by step →

Create a free account, then Premium+ runs it on your own work. Continue with Google Create a free account Log in

How it works

  1. We turn your question into its key concepts, their synonyms, and several literature searches.
  2. We run them on OpenAlex (journals, preprints and conference papers, open access or not) and on arXiv, by relevance and for the last three years, and look for reviews and the most cited works.
  3. We follow the citations both ways: the works many of the closest papers cite are the founding papers a keyword search misses, and the recent papers that cite those are the ones written with newer words. Retracted papers are left out.
  4. We rank the candidates by closeness to your question, an AI screens their abstracts and says why each one is kept, and we keep 15 to 30 papers, founding and recent, preprints and unreviewed deposits marked as such.
  5. We read the conclusions of the papers whose full text is open, then write a synthesis by approach, a comparison table, where the papers agree and disagree, what they leave open, and the numbered references. Each statement shows the sentences of the papers that support it, checked word for word.
  6. Take papers out or put others in, search again with your own queries or later to see what is new, and export the review to BibTeX, RIS, Markdown, LaTeX or Zotero.