Reference Extractor

How to Use the Reference Extractor

1. Add your paper. Press Upload .docx or .txt, or paste the text into the box. Pasting only the reference list also works.
2. Read the four figures. References, In-text citations, DOIs and Cross-check flags summarize what was found.
3. Copy what you need. Extracted references can be copied or downloaded as references.txt, and the DOIs found can be copied one per line.
4. Review the cross-check. Citations with no matching entry, and entries that are never cited, are listed so you can fix the text or the list before you submit.

How the Extractor Reads Your Paper

• The reference section. It looks for a line that is only References, Reference list, Bibliography, Works Cited, Works Consulted, Literature Cited, Sources or Sources Cited, or the usual equivalent in another of the site's languages (Referencias, Literaturverzeichnis, Références, Bibliografia, Список литературы, 参考文献 and so on), using the last such line in the document. The list runs to the end or to a heading such as Appendix, Tables, Figures, Notes or Acknowledgments.
• The entries. Lines inside the section are split into entries: blank lines separate them, and wrapped lines of one entry are rejoined. Fragments under 15 characters are dropped.
• The in-text citations. Everything above the heading is scanned for parenthetical citations containing a year, such as (Smith, 2020) or (Smith & Doe, 2021; Brown, 2018), and narrative ones such as Jones et al. (2019). Each is reduced to a first-author surname and a year.
• The DOIs. Every DOI anywhere in the document is collected once, in order of appearance.

Worked Example

Choose Try a sample. The paper has four references and cites (Smith, 2020), (Smith & Doe, 2021), (Brown, 2018) and Jones et al. (2019). The extractor reports 4 references, 4 in-text citations, 1 DOI and 2 cross-check flags:

• Missing from the list: (Smith & Doe, 2021). The list has Smith, J. (2020), but no Smith entry dated 2021, so the citation has no match.
• Never cited: Taylor, R. (2017). The forgotten study. The text does not mention Taylor.
• DOI found: 10.1000/sleep.2020, from the first reference.

The match is on first-author surname and year only, so a citation to the right author with the wrong year, as here, shows up as missing, and the list entry for that author may show as never cited. A letter after the year (2020a) is ignored when matching.

Limits to Know

• Author–date lists with the year in brackets for the cross-check. APA and Harvard entries, written Smith, J. (2020), match. Chicago author-date entries put the year after a full stop, MLA parentheticals (Smith 42) carry no year, and numbered styles carry no names, so those cannot be matched; use the Numbered Citation Order Checker for numbered lists. The reference list itself is still extracted.
• The document body only. From a .docx the extractor reads the main text. Footnotes, endnotes, headers and text boxes are not read, and older .doc files are not supported. Deleted text from tracked changes is skipped and inserted text is read, so you get the document as it looks with all changes accepted.
• Surnames with several parts. Names such as van der Berg or de la Cruz may not match between text and list.
• A heading is needed for the cross-check. If no heading is found but most of the pasted text looks like references, it is treated as a bare list: entries and DOIs are extracted, and nothing is flagged.
• Not a style check. The extractor does not judge formatting. Run the extracted list through the APA Citation Checker.

Which Tool Do You Need?

• Get the references out of a paper and cross-check them: this page.
• Count the citations in your text: Citation Counter.
• Find repeated entries in the list: Duplicate Citation Finder.
• Find the DOI of a PDF: PDF Citation Generator. Find missing DOIs in a list: DOI Finder.

Frequently Asked Questions

How do I extract references from a Word document?

Upload the .docx, or paste its text, and the extractor finds the References, Bibliography or Works Cited heading (or its equivalent in Spanish, German, French and the site's other languages) and splits everything under it into separate entries. Copy them or download them as a text file. The document needs one of those headings on a line of its own.

Is my document uploaded to a server?

No. A .docx file is a ZIP archive, and the extractor opens it in your browser using the browser’s built-in decompression. Nothing is sent anywhere, and no account is needed.

Which file types work?

Word .docx files, plain .txt files and pasted text. Older binary .doc files are not supported: open them in Word or Google Docs and save as .docx. For a PDF, use the PDF Citation Generator, which finds the DOI inside the file and cites it from publisher metadata.

How does the citation cross-check work?

Each in-text citation is reduced to the first author’s surname and the year, and each reference entry to its first word and the first (year) it contains. A citation with no matching entry is flagged as missing from the list, and an entry with no matching citation is flagged as never cited.

Why does it say a citation is missing when the source is in my list?

Usually the year differs between text and list, the first author is named differently (a two-part surname, a group author written another way), or the citation is in a style the matcher cannot read. Compare the flagged citation with the entry by eye; the check is advisory.

Can I turn the extracted references into formatted citations?

Copy the DOIs and paste them into the citation generator one per line to cite each in your chosen style. Entries without a DOI have to be searched by title or entered by hand.