Free tools for evidence synthesis

Tools

Small, single-purpose tools for the mechanical parts of a review — the merging, matching and checking that otherwise absorb the most effort. Each one runs in your browser, needs no account, and is free to use.

The tools are grouped by where they sit in a review, with a note on each about the situation it is meant for. Two of them talk to public bibliographic databases; the rest never send anything anywhere — the badge on each row says which.

Assembling the search

WoS to RIS Merge several Web of Science exports into one RIS file, with duplicates dropped and every parsing decision listed. Reach for it whenyour search returned more records than one export allows, and you are left with savedrecs (1) through savedrecs (6) to get into Zotero or Covidence as a single set. .txt · tab-delimited · .ris → .ris In-browser RIS Compare Compare two sets of references and export what is unique to each, or what they share. Reach for it whenyou re-run a search before publication and need to say exactly which records are new — or a reviewer asks what changed between the protocol search and the final one. .ris × 2 → .ris · .csv In-browser

Before screening

Deduplicator Match records across Scopus and Web of Science, keep the fuller copy of each study, and get PRISMA identification counts with a full audit log. Reach for it whenyou have searched more than one database and need one clean set to screen — plus the numbers the PRISMA flow diagram asks for, and a record of what was removed and why. .csv · .xlsx · .txt → .ris · .csv In-browser

Filling in and checking references

Reference Enricher Turn a plain reference list into full records with abstracts, each match scored against what you pasted so you can see which rows to check. Reach for it whenyou inherit a reference list — from a colleague, an older review, or a bibliography with no DOIs — and need complete records with abstracts before you can screen or import them. pasted refs · DOIs · .txt · .csv → .ris · .csv Uses APIs Reference Verifier Check a reference list entry by entry: does the work exist, is the DOI real, and which citations can no database find? Every lookup is logged. Reach for it whena manuscript, student submission or AI-drafted section arrives with a reference list, and you need to know which citations point to something that actually exists before the claim rests on them. pasted references → .ris · .csv Uses APIs

Analysing and synthesising

PDF Data Extractor Define the fields you need and have Claude or ChatGPT read up to 60 PDFs one at a time, with a live log, a copyable results table and CSV export. Reach for it whenyou have the full texts of your included studies and need a first-pass extraction table — study design, sample, outcomes, findings — to check against the papers rather than build from scratch. .pdf × 60 → table · .csv Your API key Abstract Coder Read each abstract, highlight passages and code them, then build categories and themes, write analytical memos, and pull up every excerpt behind a theme. Reach for it whenyour included studies are in a Zotero collection and you need to code what the abstracts say, see which records support each theme, and export a codebook, an excerpt table and a thematic report. .csv (Zotero or any) → .csv · .md · .html · .json In-browser

Where your files go

Tools marked In-browser read your files locally and send nothing anywhere. The two marked Uses APIs send only the reference text you paste, to Crossref, Europe PMC, OpenAlex and Semantic Scholar. The PDF Data Extractor sends your PDFs directly to Anthropic or OpenAI, using your own key.

What they will not do

None of these tools decides anything for you. They match, merge and look up; whether a record is a duplicate, or a missing reference is fabricated, remains a call you make and can overrule in the interface.

Keeping the trail

Each tool exports an audit trail alongside the records — the matching rules used, what was removed, and where every override was made — so the search can be written up and repeated.