4 ms·
NLP analysis of Pride and Prejudice
- minimaxir 10y agoHmm, why isn't the cell output included? Additionally, Jupyter Notebooks on GitHub do not render; instead, you should visit the site of the original post: http://www.cytora.com/insights/2016/11/30/natural-language-processing-in-10-lines-of-code-part-1 http://www.cytora.com/insights/2016/11/30/natural-language-p...
- bunderbunder 10y agoThey do if you don't strip the results before publishing, which seems to be what happened here.
- minimaxir 10y agoFrom reading the commands, the cells were never even executed.
- hribo 10y agoCell output is now included in repository. Thanks for the feedback.
- joemi 10y agoThat original post on cytora seems quite a bit better to me than the github link since it gives more context. Thanks!
- Xeoncross 10y agoSpacy is a great library and tutorials like this give a clear and simple path for testing it out. They even included a function for reading a file, rather than assuming the audience are all python programers. (Not that reading a file is hard, but it's an extra few minutes as you google how to read a file in X language.)
- mendelbot 10y agoCheck out textacy, a higher-level text library built on top of spaCy, for better keywords extraction: https://github.com/chartbeat-labs/textacy https://github.com/chartbeat-labs/textacy https://textacy.readthedocs.io/en/latest/api_reference.html#textacy.keyterms.textrank https://textacy.readthedocs.io/en/latest/api_reference.html#...
- wodenokoto 10y agoThey forgot to run the notebook before uploading the file to github, so there are no results. What a shame. Anyway, this is the 2nd part of a 3 part tutorial. Full setup instructions and the 2 other parts are available here: https://github.com/cytora/pycon-nlp-in-10-lines https://github.com/cytora/pycon-nlp-in-10-lines
- czep 10y agoSpacy is awesome. The API is clean and powerful. I have been using it heavily over the past few months, using it almost exclusively for feature extraction now. I'm currently working on extracting subject-verb-object tuples, which is amazingly easy to do, because I am finding these to be much more powerful than unigrams or ngrams for classification. Named Entity extraction in Spacy is another killer feature. It's been instrumental in some fraud detection work I've been doing as well.
- Xeoncross 10y ago> working on extracting subject-verb-object tuples, which is amazingly easy to do Can you share any of your work or give examples?
- czep 10y agoThis is not my work, but will show you how it's done: https://nicschrading.com/project/Intro-to-NLP-with-spaCy/ https://nicschrading.com/project/Intro-to-NLP-with-spaCy/