Skip to contents


Removing duplicate references obtained from different databases is an essential step when conducting and updating systematic literature reviews. ASySD (pronounced “assist”) is a tool to automatically identify and remove duplicate records.

Shiny application

This tool is available both as an R package and a user-friendly shiny application.


Please check out the new ASySD tutorial on youtube. We would like to extend a huge thank you to ESMARConf for giving us a platform to showcase ASySD within the evidence synthesis community!

Tool performance

An evaluation of ASySD’s performance versus other automated deduplication tools is available here


{r} install.packages("devtools") devtools::install_github("camaradesuk/ASySD")


If loading with the ASySD package functions, the required fields will be automatically created based on the input data. If metadata columns are missing, they will be automatically created by ASySD and set to NA. For optimal performance, we recommend having as much metadata as possible.

If using a dataframe, you should ideally have the following column names:

Name Definition
author The author(s) of the publication
year The year the publication was published
journal The name of the journal in which the publication appeared
doi The Digital Object Identifier (DOI) assigned to the publication
title The title of the publication
pages The page numbers of the publication
volume The volume number of the publication (if applicable)
number The issue number of the publication (if applicable)
abstract Abstract of publication
record_id A unique identifier for the publication. If this is not obtained from the citation file, ASySD will genereate an id for each citation based on row numbers.
isbn The International Standard Book Number (ISBN) assigned to the publication (if applicable). If unavailable, the International Standard Serial Number can be used here instead (ISSN).
label (optional) A label or tag assigned to the publication (if applicable) - for example, new search or old search
source (optional) The source or database from which the publication was obtained - for example wos, embase, pubmed, scopus

Automatically deduplicate citation data

```{r} # load citations citation_data <- load_search(filepath, method=“endnote”)


dedup_citations <- dedup_citations(citation_data)

get unique citation dataframe

unique_citations <- dedup_citations$unique