ScientoPy

ScientoPy is a open-source Python based scientometric analysis tool. It has the following main characteristics:

• Import Clarivate Web of Science (WoS) and Scopus data sets (recursive folder scan)
• Parallel, multi-core preprocessing pipeline (Polars CSV reader + process-pool row normalization)
• Apache Parquet as the canonical on-disk format — ~10× smaller and ~20× faster to load than CSV
• Filter publications by document type
• Merge WoS and Scopus data set based on a field tags correlation table
• Find and remove duplicated documents
• H-index extraction for the analyzed topics.
• Country and institution extraction from author affiliations
• Top authors, countries, or institutions based on first document's authors or all document's authors
• Preprocessing brief graph and report table
• Top topics and specific topics analysis
• Wildcard topics search
• Trending topics using the top average growth rate (AGR)
• Five different visualization graphs: bar, bar trends, timeline, evolution, and word cloud
• Automatic BibTeX bibliography generation from LaTeX documents using paper EIDs as cite keys
• On-demand CSV export (Scopus or WoS field style) via exportPapers.py or the GUI Export tab
• Graphical user interface with five tabs (Pre-processing, Analysis, Results, Extended Results, Export)

Download Pre-built Releases

Download the latest release for your platform from: https://github.com/jpruiz84/ScientoPy/releases

Platform	File	Instructions
Windows (x64)	ScientoPy-windows-x64.zip	Extract and run ScientoPyGui.exe
Linux (x64)	ScientoPy-linux-x64.tar.gz	Extract, then chmod +x ScientoPyGui/ScientoPyGui && ./ScientoPyGui/ScientoPyGui
macOS (Apple Silicon)	ScientoPy-macos-arm64.zip	Extract, then right-click > Open (or xattr -cr ScientoPyGui/) to bypass Gatekeeper

For detailed instructions about the ScientoPy Graphic User Interface, see the user manual: Manual/ScientoPyGui_user_manual.pdf

Run ScientoPy from source (Linux, macOS, Windows)

To clone directly the last version from the repository run the following git command:

git clone https://github.com/jpruiz84/ScientoPy

Install prerequisites

Requirements: Python 3.9 or higher.

1. On Ubuntu/Debian, install Python3 and tkinter:

   sudo apt-get install python3 python3-tk python3-pip python3-pil python3-pil.imagetk
   

2. Create a virtual environment and install dependencies:

   python3 -m venv venv
   source venv/bin/activate        # Linux/macOS
   # venv\Scripts\activate          # Windows
   
   pip install --upgrade pip
   pip install -r requirements.txt
   

The bibliometric dataset

To download a custom dataset refer to the user manual: Manual/ScientoPy_user_manual.pdf

In this repo we include an example dataset that was donwloaded using: "Bluetooth low energy" as search criteria

Running the ScientoPy Graphical User Interface GUI

To run ScientoPy operations from the GUI, execute the following command:

python3 ScientoPyGui.py

Running the ScientoPy from console scripts

This section describes the ScientoPy scripts to preprocess and analyze the bibliometric dataset.

Preprocessing

First we need to preprocess the downloaded dataset. This preprocess merges all the downloaded files from the input folder into a single file. Also, this process removes the duplicated documents. To preprocess the example dataset (“Bluetooth low energy” located in dataInExample) run this command inside ScientoPy folder:

python3 preProcess.py dataInExample

The input folder is scanned recursively, so Scopus and WoS exports can be organized in sub-folders. The pipeline is multi-core (parallel Polars CSV reader + per-file row normalization) and prints its progress as four numbered steps:

[1/4] Loading papers
[2/4] Disambiguating Scopus author names
[3/4] Removing duplicates
[4/4] Saving preprocessed corpus

Then, inside the folder ScientoPy/dataPre you will find the following files:

• papersPreprocessed.parquet: canonical preprocessed corpus stored as Apache Parquet (Zstd-compressed). Replaces the previous papersPreprocessed.csv. Used as input by every other ScientoPy script. Typically 5-15× smaller than the equivalent CSV and loads in well under a second even for 300 k-paper datasets. To obtain a CSV copy, use exportPapers.py (see below).

• PreprocessedBrief.csv: this file briefs the pre-process statics results, such as duplicated papers removed, types of documents, and others.

To find more options of the preprocessing script you can run:

python3 preProcess.py -h

Extract the top topics

With this script you can extract the top topics of a selected criterion. The ScientoPy criterion are described bellow:

• author: Authors last name and first name initial
• sourceTitle: Publication or journal name
• subject: Research areas, only from WoS documents
• authorKeywords: Author keywords
• indexKeywords: Keywords generated by the index, from WoS Keyword Plus, and from Scopus Indexed keywords
• bothKeywords: AuthorKeywords and indexKeywords are used for this search
• abstract: Document abstract, for use with pre-defined topics and asterisk wildcard
• documentType: Type of document
• dataBase: Database where the document was extracted (WoS or Scopus)
• country: Country extracted from authors affiliations
• institution: Institution extracted from authors affiliations
• institutionWithCountry: Institution with country extracted from authors affiliations

For example, to find the top author keywords you can run this script:

python3 scientoPy.py -c authorKeywords

This will generate a list with the top 10 topics on the selected criterion (in this case authorKeywords), with the number of documents per topic, and the h-index associated to each one. Also, this script graphs the evolution of each topic per year, and saves the quantitative results on the folder ScientoPy/results.

This script have more options like, save the plot on a file, or increase the number of topic results. For more information you can run:

python3 scientoPy.py -h

Analyze custom topics inside a criterion

If you want to make an analysis of custom topics, such as the two selected countries papers evolution, you can use the scientoPy.py script, with the option -t, to specify the topics:

python3 scientoPy.py -c country -t "United States; Brazil"

You can analyze any topic in any criterion. Put the topics on the -t argument. Divide the topics with the ;. Also, you can integrate two or more topics in one, by dividing it with ,. This is very useful for abbreviations and plural singulars, for example:

python3 scientoPy.py -c authorKeywords -t \
"WSN, Wireless sensor network, Wireless sensor networks; RFID, RADIO FREQUENCY IDENTIFICATION"

Note: The command is very long, for that reason the command was divided by \. If you have problems in Windows, remove the "" and put the command in one single line.

Asterisk (*) wildcard

You can use the asterisk wildcard to find phrases or words which starts or ends with the letters that you have inserted. For example, if you want to find “device”, “devices”, and “device integration”, enter the following command:

python3 scientoPy.py -c authorKeywords -t "device*"

ScientoPy will print the topics found for the previous search:

Topics found for device*:
"devices;device management;Device Interactions;Device objectification;Device;Device integration"

You can use this information, to analyze each specific topic found, like this:

python3 scientoPy.py -c authorKeywords -t \
"devices;device management;Device Interactions;Device objectification;Device;Device integration"

Evolution plot

Also, you can see the results with a evolution graphic (add -g evolution). This option plot the accumulative documents, average documents per year (ADY), and PDLY, for example:

python3 scientoPy.py -c authorKeywords -t \
"WSN, Wireless sensor network, Wireless sensor networks; RFID, RADIO FREQUENCY IDENTIFICATION" \
-g evolution

This script have more options like, save the plot on a file, or others. For more information you can run:

python3 scientoPy.py -h

Finding trending topics

This script finds the top trending topics based on the higher average growth rate (AGR) over the others. The AGR is calculated on two years periods.

To find the top trending topics on author keywords criterion, you can run the following script:

python3 scientoPy.py -c authorKeywords --trend --startYear 2008 --endYear 2018 \
--windowWidth 2  --agrForGraph -g evolution

This script will find the top 200 topics, then it calculates the AGR for the last 2 years (--windowWidth 2). Finally, the 200 top topics are sorted from the highest AGR in the last 2 year period to the lower. The first 3 AGR topics are filtered (they correspond to the keyword Internet of things), and the next 10 topics are garph in a evolution plot.

For more information about the AGR calculation refer to the PDF manual:

Manual/ScientoPy_user_manual.pdf

Analysis based on the previous results

ScientoPy writes a lightweight internal file with all the output documents from the last run so subsequent analyses can chain off it. For example:

python3 scientoPy.py -c country -t "Canada" --noPlot

ScientoPy writes results/lastAnalysis.parquet with every document whose authors are affiliated in Canada. Pass -r (or --previousResults) on the next call to analyze that subset:

python3 scientoPy.py -c authorKeywords -r -g bar

returns the top author keywords from papers affiliated in Canada.

python3 scientoPy.py -c country -r -g bar

returns the countries that most commonly co-publish with Canada.

Notes:

• lastAnalysis.parquet is only produced when -r / --previousResults is not used, so chained analyses do not overwrite it.
• In previous versions this file was a CSV named results/papersPreprocessed.csv, re-written on every analysis. It is now replaced by the Parquet file; analyses no longer write any CSV automatically. Use exportPapers.py (or the 5. Export GUI tab) to re-materialize a CSV on demand.

Exporting to CSV

The preprocessed corpus and the analysis results can be exported to Scopus- or WoS-style CSVs on demand with exportPapers.py:

# Full preprocessed corpus -> export/papersPreprocessed.csv (Scopus fields)
python3 exportPapers.py

# Same, but WoS-style field names
python3 exportPapers.py --format wos

# Copy the latest results/*.csv into export/results/
python3 exportPapers.py --source results

# Custom destination
python3 exportPapers.py --source preprocessed --format scopus -o my_corpus.csv

Running exportPapers.py with no flags defaults to --source preprocessed --format scopus and writes to export/papersPreprocessed.csv. For the full option list run:

python3 exportPapers.py -h

Extended results (per-document)

The detailed extended-results CSV (results/<Criterion>_extended.csv, one row per paper) is not produced automatically because it can reach hundreds of thousands of rows on large corpora. To produce it, either pass --saveExtended on the CLI:

python3 scientoPy.py -c authorKeywords --saveExtended

or pick Extended results (last analysis, per-document) in the GUI 5. Export tab after running an analysis.

All export options are also available from the GUI in the 5. Export tab.

Output files and directories

After run some ScientoPy commands or after run all the example commands by executing the script exampleGenerateGraphs.sh you will find the following folder and files structure described bellow:

• dataInExample: contains Scopus and WoS example data set for the search criteria “Bluetooth low energy” downloaded in April 2, 2026. This is the input example for preprocess script.

• dataPre: output folder for the preprocess results, and input folder for scientoPy script.

  • papersPreprocessed.parquet: canonical preprocessed corpus (Apache Parquet, Zstd-compressed). Replaces the old papersPreprocessed.csv as the input for every other ScientoPy script. Use exportPapers.py to obtain a CSV copy.

  • PreprocessedBrief.csv: preproceses brief table that shows the preprocess results related to total papers found per data base, the omitted papers, the duplicated papers count per data base, and the total number of papers per paper type (Conference paper, article, review...)

• graphs: graphs output folder for preprocess and scientoPy scripts

• Manual: folder with the pdf manual and example paper with scientoPy commands highlighted used for graph and tables generation.

• results: output folder for scientoPy result output files

  • AuthorKeywords.csv: scientoPy output file for the selected criterion (in this case authorKeywords) that shows the top topics or the custom topics with the total number of documents, the Average Growth Rate (AGR), the Average Documents per Year (ADY), the h-index, and the documents per each year.

  • AuthorKeywords_extended.csv: scientoPy output file for the selected criterion (in this case authorKeywords) that show the top or custom topics with the documents related to each one.

  • lastAnalysis.parquet: internal file with the output papers from the last scientoPy script run. Used as input when the option -r or --previousResults is passed on the next run.

• export: populated on demand by exportPapers.py. Default destinations are export/papersPreprocessed.csv (preprocessed corpus) and export/results/ (copy of analysis CSVs).

BibTeX generation

ScientoPy can automatically generate a BibTeX bibliography file from a LaTeX document. Use the paper EID (found in the EID column of the extended results file results/*_extended.csv) as the cite key in your LaTeX document:

\cite{WOS:000363238300013}

Then run the generateBibtex.py script:

python3 generateBibtex.py latexExample/article_example.tex

This will create latexExample/article_example_bibliography.bib next to the input file. The script:

• Extracts all \cite{} keys from the LaTeX document body
• Matches them against the preprocessed dataset (dataPre/papersPreprocessed.parquet)
• Generates @Article{} entries for journals/reviews and @Inproceedings{} for conference papers
• Handles author name formatting and LaTeX special character escaping

Reference the generated file in your LaTeX document:

\bibliographystyle{unsrt}
\bibliography{article_example_bibliography}

A Makefile is provided in latexExample/ that automates the full build (bibliography generation + pdflatex/bibtex compilation). Run make inside that folder.

This feature is also available from the GUI via the Generate BibTeX button on the Analysis tab.

ScientoPy graph types

ScientoPy has 5 different ways to graph the results described bellow:

Graph type	Argument	Description
Time line	-g time_line	Graphs the number of documents of each topic vs the publication year
Horizontal bars	-g bar	Graphs the total number of documents of each topic in horizontal bars
Horizontal bars trends	-g bar_trends	Graphs the total number of documents of each topic in horizontal bars, with the percentage of document published in the last years
Word cloud	-g word_cloud	Generate a word cloud based on the topic total number of publications
Evolution	-g evolution	Graphs two plots, one with the accumulative number of documents vs the publication year, and other with the average papers per year vs the percentage of documents in the last years

To see graph examples refer to the PDF manual:

Manual/ScientoPy_user_manual.pdf

How to cite ScientoPy

If you use ScientoPy in a book, paper, website, technical report, etc., please include a reference to ScientoPy.

To cite ScientoPy, use the following reference:

Ruiz-Rosero, J., Ramirez-Gonzalez, G., & Viveros-Delgado, J. (2019). Software survey: ScientoPy, a scientometric tool for topics trend analysis in scientific publications. Scientometrics, 1-24.

The bibtex entry for this is:

@Article{Ruiz-Rosero2019,
    author="Ruiz-Rosero, Juan
    and Ramirez-Gonzalez, Gustavo
    and Viveros-Delgado, Jesus",
    title="Software survey: ScientoPy, a scientometric tool for topics trend analysis in scientific publications",
    journal="Scientometrics",
    year="2019",
    month="Nov",
    day="01",
    volume="121",
    number="2",
    pages="1165--1188",
    abstract="Bibliometric analysis is growing research filed supported in different tools. Some of these tools are based on network representation or thematic analysis. Despite years of tools development, still, there is the need to support merging information from different sources and enhancing longitudinal temporal analysis as part of trending topic evolution. We carried out a new scientometric open-source tool called ScientoPy and demonstrated it in a use case for the Internet of things topic. This tool contributes to merging problems from Scopus and Clarivate Web of Science sources, extracts and represents h-index for the analysis topic, and offers a set of possibilities for temporal analysis for authors, institutions, wildcards, and trending topics using four different visualizations options. This tool enables future bibliometric analysis in different emerging fields.",
    issn="1588-2861",
    doi="10.1007/s11192-019-03213-w",
    url="https://doi.org/10.1007/s11192-019-03213-w"
}

Authors

• Juan Ruiz-Rosero - Initial work and documentation

License

This project is licensed under the MIT License - see the LICENSE.md file for details

Additional Information

• ScientoPy will not transfer any information to other networked systems unless explicitly requested by any party involved with installing or operating it

Sponsors

Free code signing on Windows provided by SignPath.io, certificate by SignPath Foundation