SpaCy pipeline and models for scientific and biomedical text processing.
Process scientific documents with custom tokenizers, POS tagging, and NER models.
scispacy provides custom spaCy pipelines and models tailored for scientific and biomedical documents. It includes a custom rule-based tokenizer, POS tagger, syntactic parser, and entity span detection models.
The library offers specialized pipelines like en_core_sci_sm, md, lg, and scibert, along with NER models trained on corpora such as CRAFT, JNLPBA, BC5CDR, and BIONLP13CG. It also provides an AbbreviationDetector and an EntityLinker component to link entities to knowledge bases like UMLS, MeSH, RxNorm, GO, and HPO.
It suits bioinformaticians and NLP engineers working with biomedical literature who need specialized tokenization and entity recognition. It does not suit general-purpose English text processing outside scientific domains.
Includes a custom tokenizer adding tokenization rules specifically for scientific documents.
Offers models trained on specialized corpora like CRAFT, JNLPBA, BC5CDR, and BIONLP13CG.
Implements the Schwartz-Hearst algorithm to extract abbreviations and long forms in text.
Links extracted entities to ontologies such as UMLS, MeSH, RxNorm, Gene Ontology, and Human Phenotype Ontology.
The tool is free to use under its open-source license. The only cost is the compute resources required to run the models yourself.
880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
You must install the core library via pip and then separately download and install a compatible model package from the provided release URLs.
Yes, using the EntityLinker component you can link entities to knowledge bases including UMLS, MeSH, RxNorm, GO, and HPO.
See Zekai first in Google