Ensemble lemmatization with the Classical Language Toolkit
Parole chiave:lemmatization, natural language processing, Latin, Classical Language Toolkit
AbstractBecause of the less-resourced nature of historical languages, non-standard solutions are often required for natural language processing tasks. This article introduces one such solution for historical-language lemmatization, that is the Ensemble lemmatizer for the Classical Language Toolkit, an open-source Python package that supports NLP research for historical languages. Ensemble lemmatization is the most recent development at CLTK in the repurposing and refactoring of an existing method designed for one task, specifically the backoff method as used for part-of-speech tagging, for use in a different task, namely lemmatization. This article argues for the benefits of ensemble lemmatization, specifically, flexible tool construction and the use of all available information to reach tagging decisions, and presents two use cases.
Articles and submissions processing charges (APC)
This journal does not charge Article Processing Charges (APC) and Article Submission Charges (ASC).
Deposit and Self-archiving policies
– Authors are allowed to upload their papers immediately after publication on limited-access institutional repositories or archives. Authors ought to include publication references (journal title, volume, issue, and pages, article DOI when available, URL to journal website or journal issue).
– Six months after publication, authors are allowed to upload their submitted manuscripts in pre-print version – but not the published version – on openly accessible archives or repositories (including personal websites and institutional personal pages and personal profiles on academic social media, etc...). It is highly recommended to include a reference to the published version.
– Five years after publication, the article is released under a CC BY SA 4.0 license and kept on the journal website. All rights revert to the author.
– Authors may purchase early open access and immediately release their published paper (200 EUR fee).