Feature Extraction

Feature extraction is the process of converting a list of texts into some kind of feature matrix.

Parameters in the config file should be under the section [feature_param].

We have currently implemented the following feature extraction methods:


Use the standard TF-IDF (Term Frequency-Inverse Document Frequency) feature extraction from SKLearn.

Gives a sparse matrix as output. Works well in combination with Naive Bayes and other fast training models (given that the features vectors are relatively wide).

See asreview.feature_extraction.Tfidf.


Feature extraction method provided by the gensim package. To use it, please install the gensim package manually:

pip install gensim

It takes relatively long to create a feature matrix with this method. However, this only has to be done once per simulation/review. The upside of this method is the dimension-reduction that generally takes place, which makes the modelling quicker.

See asreview.feature_extraction.Doc2Vec.


Feature extraction method where the average of the word embeddings are taken of words in the text, multiplied by the inverse document frequency of said words.

See asreview.feature_extraction.EmbeddingIdf.


Feature extraction method for LSTM/RNN models.

See asreview.feature_extraction.EmbeddingLSTM


Feature extraction method based on Sentence BERT. Install the sentence_transformers package in order to use it. It is relatively slow.

pip install sentence_transformers

See asreview.feature_extraction.SBERT