REVIEW REFERENCE DUPLICATES BEFORE REMOVING THEM

Research Reference Deduplicator

Compare CSV reference lists by DOI, canonical URL or normalized title. Review duplicate candidates, choose a record to keep and export the original columns without merging conflicting DOIs.

1. Reference inventory

CSV table with title, doi, url and/or year headersOpen UTF-8 reference CSVDelimiter Also compare normalized titles (minimum 12 characters) Remove common tracking parameters for URL comparison Prefix formula-like CSV cells with an apostrophe on export

3,000 data rows · 80 columns · 500,000 characters · 30-second worker limit. Rectangular CSV with unique headers required. At least one title, doi or url header is needed; extra columns are retained.

2. Choose retained records

Load the sample or your reference table.

Chosen-record CSV preview

Matches are candidates, not verified publication identities. Review versions, corrections and editions. This tool does not query Crossref, publishers or DOI resolvers.

How to review a bibliography inventory

Export references as CSV using recognized headers: title, doi, url, year. Keep author and other metadata in additional columns.
Choose comparison rules and find candidates. DOI matching comes first, then URL, then optional title matching.
Review each duplicate group and select its retained record. Fields are not combined automatically.
Export chosen rows in original order and preserve the full review JSON if you need the grouping record.

Conservative matching rules

DOI URLs and doi: prefixes normalize to a lowercase identifier; syntax checks do not establish registration. URLs retain query values except explicitly removed tracking parameters; fragments are ignored. Titles use compatibility normalization, lowercasing and punctuation/space normalization.

Different nonempty DOIs are never grouped by title or URL, even through a record without a DOI. Different nonempty years block title-only grouping. Equal DOI or URL with differing metadata still needs manual review. Invalid metadata is flagged and excluded from that matching key; other valid keys can still identify candidates.

CSV export can prefix formula-like cells, changing those values for spreadsheet safety. Turn the option off only if you need exact unmodified cell strings. The report contains complete source records; no automatic storage is used.

Using Research Reference Deduplicator

Review duplicate publication candidates in a reference CSV without external lookups or automatic metadata merging.

Import a CSV table with title, doi, url and optionally year headers.
Inspect candidate groups and select one record per group.
Export chosen records and retain the complete review report.

Example

The sample groups a DOI/URL duplicate pair while keeping a conflicting DOI separate.

Questions & answers

Does it verify references online?

No. DOI and URL syntax are normalized locally without resolver, publisher or Crossref requests. Valid-looking metadata does not prove a publication exists.

Can conflicting DOIs be merged?

Different nonempty DOIs are never grouped through URL or title matching, including indirect bridges through records without DOIs. Different nonempty years block title-only matches.

Are all fields merged?

No. Select one original record per group; its original columns are retained. Extra metadata is not merged or invented. Singletons remain in the output.

What inputs are supported?

Rectangular CSV with unique headers, up to 3,000 data rows, 80 columns and 500,000 characters. At least one title, doi or url header is required. Jobs stop after 30 seconds.

What is in the review report?

Complete source records, normalized comparison fields, candidate groups, conflicts and chosen record indices. Treat it as sensitive as the original; no automatic storage is used.