MatchPub2 scans the literature to retrieve papers that were handled by a journal and analyzes their fate (where they were published) and citation distribution as a function of the editorial decision made by the journal (accept, reject before review, reject after review).
As input, MatchPub2 ingests the data produced with the eJP SQL query queries/matchpub.sql in .xls format.
An alternative input is the ejp editor_track_report.xls report (using the detailed format). In this case, the decisions labels used in this report need to be mappted to the three fundamental decisions -- accept, rejected before review, rejected after review -- by updating accordingly the regular expressions provided in decision.py. The naming and the order of the columns of the editor_track_report.xls can be customized in descriptions.ejp_editor_track_report as described in the modules' documentation.
By default, content of published papers is retrieved from EuropePMC (but see below on how to use PubMed instead).
Citations data are obtained from Scopus.
Clone this repository.
Update .env.example with your user id (id -u) and group id (id -g), your Scopus API_Key (register at https://dev.elsevier.com/).
Install docker and docker-compose (https://www.docker.com/get-started).
Build the application:
docker-compose build
Download the eJP report into the data/ directory (or /data from within the container since data/ is 'bind mounted' (https://docs.docker.com/storage/bind-mounts/) to the container's /data volume).
Run a scan from the command line within the docker container:
docker-compose run --rm matchpub bash
python -m src.scan /data/to/ejp_report.xls> /results/to/result/> # use --no_citations to prevent inclusion of citation data.
The default search engine is EuropePMC. To use PubMed instead, use the flag --use_pubmed. In our experience, PubMed is slower and does not lead to higher recall rates.
To obtain debug-level information run the scan with -D option.
To prevent inclusion of Scopus citation data, use the --no_citations flag.
In addition to the specified <result>.xlsx file, MatchPub will save a <result>-not-found.xlsx> file with the list of papers that could not be matched. Graphical reports will be saved in /reports.
To run the interactive visualization in a Jupyter notebook:
docker-compose up
# use link provided as output: http://127.0.0.1:8888/?token=<...> to access the notebook in the browser
Shut down with Ctrl-C
Some settings can be changed in src/config.py.
To scan only preprints, set preprint_inclusion to PreprintInclusion.ONLY_PREPRINT.
To avoid the inclusion of citation data, set include_citations to False or invoke src.scan with the --no_citations option.
A customizable description of the format of the input file can be specified with input_description_file.
Descriptions of input files and their documentation are provided in src/description.py.
This reports provides an overview of the 'quality' of the scan. It main purpose is visualize the fraction of papers that could be retrieved.
This reports shows the journals in which rejected papers were ultimately published. The size of the sectors is proportional to the number of papers.
This reports plots Scopus citation data for papers accepted, rejected before review and rejected after review. With the interactive HTML version of this report, hovering over a point displays information on the journal, doi, title and manuscript number of the paper.
This report displays citation data for papers accepted, rejected before review and rejected after review. The data is split into 150 citation bins and displayed as a histogram distribution.
This is a less serious report...
The papers of the journal (the "Grand Cru") are merged with an "Assemblage" of rejected papers that were ultimately published in the external journals to obtain a 'synthetic journal' named "Cuvee". The N (default n_top=3) largest destination external journals shown in the 'manuscript fate' report are used.
The impact of this 'merger' is visualized by comparing the disribution of the 'Grand Cru', 'Cuvee' and 'Assemblage'.
