ExChain

Very simple Markov Chain written in Elixir. This is my toy-project, so I can learn Elixir.

If you have experience with Elixir and find improvements, please let me know, or simply write a comment in the appropiate line. All suggestions are more than welcome.

It's mostly based in the Markov Models in Ruby tutorial.

The Markov Model creates funny messages based on tweets downloaded from your Twitter Archive.

It generates messages like this.

How to run this?

The first thing you need is Elixir. If you have a Mac, you can install it using brew install elixir.

This program needs your twitter archive, so go ahead and dowload it from your settings in twitter.com.

Download the repository with git clone git@github.com:eljojo/ex_chain.git. Extract your twitter archive, and put the contents of the data/js/tweets folder into ex_chain/data.

$ cd ex_chain
$ mix deps.get # get the dependencies using mix, Elixir's rake. Analog to bundle install.
$ iex -S mix # analog to irb or rails console

Inside iex, you can generate new messages by typing:

iex(1)> {:ok, model} = ExChain.MarkovModel.start_link
{:ok, #PID<0.126.0>}
iex(2)> ExChain.MarkovModel.populate_model(model, ExChain.FileDatasource.get_data)
{:ok}
iex(3)> ExChain.SentenceGenerator.create_filtered_sentence(model)
{:ok, "I'm not a Food Social Network?", 0.4668037713169559, 4}
iex(4)> ExChain.SentenceGenerator.create_filtered_sentence(model)
{:ok, "I am in churros.", 0.30839889769910056, 2}
iex(5)> ExChain.SentenceGenerator.create_filtered_sentence(model)
{:ok, "I'm never sleeping again.", 0.3546596799639396, 14}
iex(6)> ExChain.SentenceGenerator.create_sentence(model)
{"future: scumbag apple: makes you release the happiness hormone.",
 0.8334241103848947}
iex(7)> ExChain.SentenceGenerator.complete_sentence(model, "i love")
{"i love the smell of shipping technical debt", 0.6685185185185185}

Where's the code?

The meat and potatos are located in just three files:

File	what it does
lib/ex_chain/file_datasource.ex	This file parses the tweets located in the data folder.
lib/ex_chain/markov_model.ex	This file is in charge of holding the state of the "Markov Model".
lib/ex_chain/sentence_generator.ex	This file generates the sentences using the Markov Model.

CLI

ExChain also has a CLI interface. To use it, you first need to compile the application into an escript (Erlang VM) binary.

$ mix escript.build
$ ./ex_chain generate_sentence
I think currywurst is the best time to wake up.

Suggestions

Please send me all your suggestions to @eljojo or directly to my email jojo at eljojo dot net.