chistes-nlp

Introduction

In this workshop different NLP techniques are explored to solve a supervised categorization problem: given a collection of Spanish jokes and their tagged categories, use a ML model to predict the category of a new joke.

This workshop is intended to show the following:

• Classical and modern NLP techniques.
• Python libraries used to work with natural language.
• The specific problems working with Spanish language.
• The difficulties using a confusing domain: jokes.

This workshop takes part in:

• PyDay BCN 2018, November 24th; Barcelona
• PyConEs 2019, October 4rd-6th; Alicante

Installation

It is recommended to install the requested libraries before attending the workshop, so you don't need to rely in conference's WiFi. There are 2 recommended ways: using pipenv or just using pip to install the requirements.

Using pipenv

Just do the following:

pipenv install
pipenv shell
python -m spacy download es
jupyter notebook

First command is used to install the dependencies; 2nd, opens a shell in the generated virtual environment; 3rd, downloads Spanish data needed by spacy; 4th, opens jupyter in your navigator.

If there was no problems 🤞, you should see jupyter on your browser: just open chistes-nlp.ipynb file.

Using pip

You can install the required libraries directly in your computer, using the following commands:

pip install -r requirements.txt
python3 -m spacy download es

But it should be better to use a virtual environment, like virtualenv. Refer to its documentation for more information, or just use pipenv (previous section).

Contents

You may only need to see these two:

• slides/ : slides of the introduction to NLP
• chistes-nlp.ipynb : notebook to be used in the workshop

Other files:

• chistes-nlp-all-executed.ipynb : notebook with all cells executed, in case you can't execute yourself
• data/ : 2k Spanish jokes dataset
• Pipfile, Pipfile.lock, requirements.txt : dependencies' files