Skip to main content

Citation style classifier

Project description

Citation style classifier

Citation style classifier can automatically infer citation style from a reference string. The classifier is a Logistic Regression model trained on 90,000 reference strings. The following citation styles are supported by default:

  • acm-sig-proceedings
  • american-chemical-society
  • american-chemical-society-with-titles
  • american-institute-of-physics
  • american-sociological-association
  • apa
  • bmc-bioinformatics
  • chicago-author-date
  • elsevier-without-titles
  • elsevier-with-titles
  • harvard3
  • ieee
  • iso690-author-date-en
  • modern-language-association
  • springer-basic-author-date
  • springer-lecture-notes-in-computer-science
  • vancouver
  • unknown

The package contains the training data, the classification model, and the code for feature extraction, selection, training and prediction.


    pip3 install styleclass


From command line:

    styleclass_classify -r "reference string"
    styleclass_classify -i /file/with/reference/strings/one/per/line -o /output/file

In Python code:

    from styleclass.classify import classify
    from styleclass.train import get_default_model

    model = get_default_model()
    prediction = classify("reference string", *model)
    prediction = classify(["reference string #1", "reference string #2", "reference string #3"], *model)


Styleclass package contains two datasets: training set and test set. Each of them contains a sample of 5,000 DOIs formatted in 17 citation styles (listed above), which gives 85,000 reference strings. Both datasets were generated automatically using Crossref REST API.

A new dataset can be generated using the script styleclass_generate_dataset.


The default model was trained on the training dataset. Before the training, the dataset was cleaned and enriched with random noise. 5,000 strings with "unknown" style were also generated and added to the dataset.

Script styleclass_train_model can be used to train a new model. This is useful especially when you need to operate of a different set of citation styles than our default. The script prepares the data for training in the same was as was done for training of the default model.


styleclass_evaluate script can be used to evaluate exisitng model on a test set, in terms of accuracy.

The accuracy of the default model estimated on our test set is 95%.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for styleclass, version 0.0.5
Filename, size File type Python version Upload date Hashes
Filename, size styleclass-0.0.5-py3-none-any.whl (5.9 MB) File type Wheel Python version py3 Upload date Hashes View
Filename, size styleclass-0.0.5.tar.gz (5.9 MB) File type Source Python version None Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page