List of extensions for ASReview LAB

[Rensvandeschoot]

ASReview has extensive support for (community-maintained) extensions. In this discussion thread, we list all such extensions. Is an extension missing in this list? Leave a message below. Would you like to read more about developing a new extension? Read Create an Extension in the documentation. For more information about the use of extensions, see Extensions in ASReview LAB.

Metrics, analytics, and plots

ASReview-datatools can be used for describing basic properties of a dataset (e.g., number of papers, number of inclusions, the amount of missing data and duplicates), converting file formats via the command line, and cleaning your (input) data by removing duplicate records. Maintained by the ASReview development team.

ASReview-wordcloud can be used to create a visual impression of the contents of the titles and abstracts in a dataset via a word cloud. Maintained by the ASReview development team.

ASReview-insights can be used to extract performance metrics from the .asreview file to evaluate the performance of simulation studies. The metrics can be used to create several plots which can be adjusted via the plotting-API. Maintained by the ASReview development team.

Model extensions

vocab-extractor can be used to extract the vocabulary and vector matrices from the .asreview file. For TF-IDF it saves the matrix and the vocabulary to pickle and JSON, respectively, and for doc2vec it stores the entire doc2vec model. Maintained by jteijema.

multilingual-feature-extractor implements a multilingual feature extractor algorithm to ASReview LAB. This algorithm allows for the usage of textual data in multiple languages: Arabic, Chinese, Dutch, English, French, German, Italian, Korean, Polish, Portuguese, Russian, Spanish, and Turkish. Maintained by jteijema.

17-layer-CNN-classifier implements a 17-layer deep convolutional neural network (CNN) model for use in ASReview. Maintained by jteijema.

model-switcher implements a model that switches between two models during simulation runtime. It can be helpful to investigate the performance when later stages of data classification require a different model than the first stage. Maintained by jteijema.

CNN-HPO implements a model consisting of two separate classifiers for use during simulation mode. The first X amount of iterations (default = 500) are run with a Naïve Bayes model. After the switching, a switch to a CNN is made. Immediately at this switching point, after every 150 iterations, hyperparameter optimization is conducted to find the most suitable CNN architecture for the current iteration. Maintained by BartJanBoverhof.

wide-doc2vec implements doc2vec with a wider vector size (120).

asreview-hyperopt can be used to optimize the hyperparameters of the models in ASReview. Maintained by the ASReview development team.

Dataset extensions

ASReview-COVID19 adds the CORD19 database and the COVID19 preprints dataset to ASReview LAB. Note: This extension is deprecated. It still works for version 0.x of ASReview but datasets are no longer updated.

Misc

irr computes the inter-rater reliability. It generates a report with the number of abstracts reviewed by each reviewer, the number flagged as relevant, the number left unreviewed, the Kappa statistic, and the number of abstracts for which researchers A and B disagreed (e.g., A said relevant, B said irrelevant). Maintained by langtonhugh.

XREF2CSV-tool written in NodeJS converts XREF-XML files to CSV files that can be imported to ASreview. Maintained by erikvullings.

[jteijema]

ASReviews' Makita (MAKe IT Automatic) is a workflow generator for simulation studies using the command line interface of ASReview LAB. Makita can be used to simplify your own research by enabling you to effortlessly generate the framework and code for your simulation study.

[Rensvandeschoot]

The ASReview Notes Export Extension is an extension that can be used to export notes from the ASReview GUI while exporting the labeled or partially labeled dataset as .csv file. It solves discussion #1220

[rohitgarud]

The extension is no longer needed after the latest release of ASReview 1.2 as it is now a core functionality. Look for exported_notes_# in the exported dataset.

[rohitgarud]

Asreview-Ensemble-Classifiers extension adds a new set of classifiers by ensembling different basic classifiers such as Naive Bayes (NB), Logistic Regression (LR) and Random Forest (RF). A simulation study performed using these classifiers is presented at #1334

[rohitgarud]

Asreview-SimilarityClassifier extension adds a set of classifiers based on the similarity between the features of the relevant records and the unlabelled records. A simulation study performed using the similarity classifier is presented at #1371

[george-gca]

AI Conference Papers is a set of unlabeled data (it doesn't have any paper flagged as useful/not useful) regarding some of the most influential conferences in the AI scene over the last years.

It contains readily importable tsv files containing (title, abstract, url, authors) for the following conferences: AAAI, ACL, COLING, CVPR, EACL, ECCV, EMNLP, Findings of the Association for Computational Linguistics, ICCV, ICLR, ICML, IJCAI, IJCNLP, KDD, NAACL, NeurIPS, SIGCHI, SIGDIAL, SIGGRAPH, TACL, and WACV.

[george-gca]

ASReview Top2Vec is based on the ASReview Semantic Clustering, but uses Top2Vec underneath. It automatically detects the number of topics it can divide the documents and group them. It can also provide more information about the topics, like most relevant words for each topic. I ditched the visualization inside for one using Tensorflow's Embedding Projector, since it is more flexible, and allows visualizing the documents and its metadata, also changing the dimensionality reduction algorithm to use.

[jteijema]

Cool! I'll definitely check this out.