SciSpace Academy
Tutorial

How to Write Better Prompts for SciSpace Deep Review

Rohan TondulkarEnglishStart here

TL;DR

1-min read

The quality of an AI literature review depends heavily on the question you ask. In this hands-on tutorial, Rohan Tondulkar, Research Scientist III at SciSpace who leads the SciSpace Research team, explains how to frame prompts for the Deep Review feature and how to work with the results. It is aimed at PhD and postgraduate researchers, early-career academics, research assistants and faculty who feel overwhelmed by the volume of literature.

Key takeaways

  • Different prompt types suit different goals, such as comparing studies, tracking recent trends or understanding methods.
  • Side-by-side examples show how weak and strong prompts differ; a good prompt states a clear goal, names the population, concept or method, and sets limits such as time period or field.
  • Choosing the right Deep Review mode and narrowing by year, field or keywords keeps results focused.
  • AI-generated summaries should be read critically, using filters and paper comparisons to check claims and spot research gaps.
  • Reviews can be saved, edited and reused in a thesis, journal paper or grant proposal.
  • Summaries work best as a map of the literature: use them to decide which papers to read in full, verify every claim against the source and write the final synthesis in your own words.
Apply it

To apply this, rewrite your next Deep Review query with a specific goal, scope and timeframe, compare the results with your original broad query, and refine from there.

How to do it

  1. Match prompt type to your goal

    Choose a prompt type that fits your goal: comparing studies, finding recent trends or understanding methods. Good SciSpace Deep Review prompts start with a clear purpose.

  2. Turn weak prompts into strong ones

    Compare weak and strong prompts side by side and rewrite yours to be more specific. A stronger question produces a better review.

  3. Choose a mode and narrow scope

    Select a Deep Review mode and narrow the scope by year, field or keywords. Focused scope gives more relevant results.

  4. Interpret summaries and spot gaps

    Read the AI summaries, apply filters, compare papers and spot research gaps. Check summaries against the papers before relying on them.

  5. Save and reuse your reviews

    Save, edit and reuse reviews in your thesis, paper or grant proposal. Reusing work saves time across projects.

Watch the tutorial44 min

How to Write Better Prompts for SciSpace Deep ReviewPlay video: How to Write Better Prompts for SciSpace Deep Review

Chapters

  1. Welcome and speaker introduction
  2. What an AI literature review tool should do
  3. How does AI semantic search find research papers?
  4. Standard vs High Quality vs Deep Review in SciSpace
  5. How Deep Review does backward and forward citation searching
  6. How to search for relevant literature in SciSpace
  7. How to show only papers from Q1 journals and top conferences
  8. Chat with PDF from search results: summaries and follow-up questions
  9. Sort, export to CSV, journal and Boolean-style filters
  10. High Quality mode and adding columns to compare papers
  11. SciSpace Deep Review walkthrough: running your first query
  12. How to answer Deep Review's clarifying questions
  13. Deep Review results: 300 papers and a cited literature report
  14. How to find only review papers from the last 5 years with a prompt
  15. How to search papers from a specific journal like Nature
  16. How to find clinical trials with 1,000+ participants using AI
  17. How to limit Deep Review to papers with PDFs
  18. Save the report to a notebook and switch citation styles
  19. Skip the clarifying questions by typing run
  20. What is the first question to ask a research paper?
  21. Deep Review output: categorized report, paper summaries and AI podcast
  22. Deep Review prompt templates for better literature reviews

Try this on SciSpace

Frequently asked questions

What makes a good prompt for an AI literature review?

A good prompt states a clear goal, names the population, concept or method you care about, and sets boundaries such as time period or field. Asking how recent studies compare two methods for a specific problem gives more useful results than a single broad keyword. Refine the prompt after reading the first set of results.

Can AI-generated literature summaries be used directly in a thesis?

They are best treated as a map of the literature rather than finished text. Use them to decide which papers to read in full, verify every claim against the original source, and write the final synthesis in your own words with proper citations. Check your university's rules on AI use and disclosure before submitting.

Related tutorials