The goal of this project is to create a random name generator that learns from a data set of existing words (e.g. English dictionary, list of names, etc.) and establishes a Follow Likelihood Matrix to determine the likelihood of each letter to be the next one in a sequence. Since each next character only depends on the current character, random name generation can then be expressed as a Markov chain based on these likelihood values, and should result in mostly pronounceable names. The model can be trained on any plain-text list of words, which may yield different results in the random name generation.
Install dependencies with pip:
pip install -r requirements.txt
This script takes three parameters, in this order:
num_names: number of names to be generated, int, optional
length: the number of characters in each name, int
model_base_name: the base name of the Markov model, without _follow, _initial, or .npy extension
Example - Generate 10 names with 6 characters each, using the model english-words-dwyl:
python3 random_name_generator.py 10 6 english-words-dwyl
Note: The model either has to be in the same directory, or the model_base_name must include the path to the model.
This script takes two parameters, in this order:
input_filename: path to the training file (plain-text list of words)
output_file_basename: the desired base name of the output Markov model, without _follow, _initial, or .npy extension
Example - Generate a model models/english-words-dwyl based on a list of words data/english-words-dwyl.txt:
python3 markov_model_builder.py data/english-words-dwyl.txt models/english-words-dwyl
This will create two files: models/english-words-dwyl_initial.npy and models/english-words-dwyl_follow.npy. For random name generation, the model_base_name will be english-words-dwyl. Note that this model already exists in this repo.
english-words-dwyl.txt: https://github.com/dwyl/english-words/blob/master/words.txt