sng 0.3.2

Generate name proposals for companies, software, etc.

• This package on PyPI

• This package on GitHub

Summary

This package can train a model that learns the “structure” of the words in a supplied text corpus. It then generates new words with a similar structure, which can be used as suggestions for naming things like companies or software.

Quickstart

Check out the Jupyter Notebook(s) in doc/notebooks/.

Documentation

• The full documentation is available online

• I also gave a lightning talk presenting the basic idea, it’s available on Youtube.

Extended summary

Naming a startup is hard.

I therefore wrote a Python package to randomly generate company name ideas.

It takes an arbitrary text as input, and then trains a recurrent neural network (RNN) on each its words, learning the structure of the text. The input text can be a simple word list (e.g. Greek or Gallic), or a chapter from a book, or just a random list of words (e.g. all Pokemon). The script then generates new random names that sound simliar to the provided list.

Literature/References

• Andrew Ng’s Deep Learning MOOC

• http://karpathy.github.io/2015/05/21/rnn-effectiveness/

• https://github.com/keras-team/keras/blob/master/examples/lstm_text_generation.py