Up until quite recently the main go to part-of-speech tagger (POS) for use in Python was the NLTK Tagger using the default Penn Treebank tagger. Another popular option was from the Stanford Log-Linear Part-Of-Speech Tagger from Stanford CoreNLP.
In May 2016 Google open sourced SyntaxNet with the humorously named English parser Parsey McParseFace. Another commercial company Spacy.io also provides an open source parser.
The development of Part-Of-Speech tagging continues in academia, but most are not at the stage where they can be easily used outside of research. How do these taggers differ? This talk will give an overview of part-of-speech tagging from a practical implementation/use-case point of view.
- What is part-of-speech tagging
- Real life performance on datasets
- How you might use POS in a project
- Compare and contrast the above mentioned taggers for performance.