v0.0.13

Fix small document split and stop words bugs

v0.0.12

This release fixes stop word removal bugs, fixes memory issues for long documents, adds a build_tokenizer attribute, and online update functions.

Solves issues #34, #31, #29, #28, #26, and #6.

Add spacy.Language as valid argument for 'spacy_pipeline'

This release allows to reuse an object from spacy.load for many different KeyphraseVectorizer objects. This release includes PR #19

Custom POS-tagger feature added

Added the options to use a custom POS-tagger, define custom stop words, and exclude certain spaCy pipeline components. This release solves issues #2 and #7.

Higher compatibility with available SpaCy pipelines

Fixed issue #11 and #10 by removing the default exclusion of certain spaCy pipeline components. This slightly slows down the keyphrase extraction process. However it grants higher compatibility to all available spaCy pipelines, including the ones that use transformers.

Added 'stop_words'=None option

Fixed #8

Add stopwords download automation

v0.0.7

Signed-off-by: Tim Schopf <[email protected]>

Change "multiprocessing" parameter to "workers" parameter

change "multiprocessing" parameter to "workers" parameter

Signed-off-by: Tim Schopf <[email protected]>

Added min_df and max_df parameters, added support for documents that have more than 1000000 characters, and limit max keyphrase length to 8 words to prevent memory issues

update scipy requirements

Signed-off-by: Tim Schopf <[email protected]>

Increased efficiency of spaCy pipeline for POS tagging

v0.0.4

v0.0.4, increased efficiency of spaCy pipeline for POS tagging + adde…