PUNchlineGenerator

Generates a pun-like variation of a user-entered phrase. Incorporates a Levenshtein distance algortithm in conjunction with the CMU Pronouncing Dictionary.

Contains:

• Edit_Distance.ipynb Jupyter Notebook file with code showing how edit distance works
• Parts_of_speech.ipynb Jupyter Notebook showing some basic part-of-speech tagging
• Pronunciation.ipynb Jupyter Notebook showing options for phonetic representations
• PUNchline.ipynb Jupyter Notebook file with the actual "pun" code
• real_words_lower.csv CSV file of "real" words and their pronunciations. (CMU Dict contains some weird words that are removed from this)
• all_words.csv CSV with words, pronunciations, and parts of speech. This includes some plurals that were not in Brown Corpus for some reason
• PG_Pun_Talk.pptx PowerPoint presentation from the PyGotham 2017 talk
• scripts/ Scripts to make setting up the enviroment easier for those who is Anaconda

Setup

If you just want to run the main code, not setup is required. These additional libraries are for supplementary material.

For those who use Anaconda:

• Clone repository locally
• From within the repo, call bash scripts/setup.sh. This should create a conda environment for you.
• Activate environment with source activate pun_generator

Otherwise, these are the requirements:

• nltk
• jupyter
• fuzzywuzzy and fuzzywuzzy[speedup] https://github.com/seatgeek/fuzzywuzzy