mit-nlp / MITIE

MITIE: library and tools for information extraction
2.92k stars 534 forks source link

enhancement: direct access to word embeddings #59

Open amn41 opened 8 years ago

amn41 commented 8 years ago

I'd like to build a simple sentence classifier using the spectral embeddings a simple average-of- bag-of-words approach.

Thinking about forking this and creating a model using dlib, but it might make sense anyway to write some functions which just return the embeddings for a list of tokens which can then be wrapped in python and other bindings. Let me know what you think

davisking commented 8 years ago

The C++ API for doing this is already straightforward. But if you wanted to contribute a Python API for it that would be cool. :)

amn41 commented 8 years ago

Cool - I'm currently making a Python API for the text_categorizer. Will submit a PR soon and we can test it out.

davisking commented 8 years ago

Sweet.

avitale commented 7 years ago

Hi, I have added a Python API for the Total Word Feature Extractor and submitted a pull request, let me know if it's ok!

davisking commented 7 years ago

Cool, yeah, I just added some comments.