Shard Detail

hailo v0.2.1

Markov chatterbot like MegaHAL
markov-chain chatbot

Install & Use

Add the following code to your project's shard.yml under:

dependencies to use in production
- OR -
development_dependencies to use in development


hailo:
  github: hailo/hailo.cr

Readme

đŸ”´ hailo.cr

License GPL 3 Travis CI

Hailo is a Markov chatterbot inspired by MegaHAL.

It is actually a port of the eponymous Hailo Perl module.

The difference being that this one's written in Crystal, consists of less code, runs faster, and uses less memory. It also drops support for multiple storage engines, sticking with SQLite only.

Differences from classic MegaHAL

  • Uses a space-efficient SQLite database to store the brain (saving 90% of the space)
  • Quite fast at training against a large corpus
  • Includes a server daemon
  • Won't choke when the number of unique tokens passes the 64k mark ;)
  • Sophisticated parser and formatter
    • Meaningfully tokenizes almost anything you'd encounter in a chatroom (Unicode, URLs, dates & times, quotes, abbreviations etc)
    • Intelligent handling of punctuation as prefix/postfix/infix tokens
    • Special consideration of fully-capitalized and partially-capitalized tokens rather than normalizing case and treating them identically
    • Outputs clean, properly capitalized sentences

Installation

First, install Crystal along with libsqlite3-dev and libreadline-dev. Then do:

git clone https://github.com/hailo/hailo.cr.git
cd hailo.cr
shards install
crystal build bin/hailo-cr.cr --release

This gives you a hailo-cr binary that you can run.

Usage

Use the command line interface (hailo-cr), or use Hailo in your code:

require "hailo"

hailo = Hailo.new("test.sqlite")
puts hailo.learn_and_reply("oh hi there")

See the API documentation for more information.

Hacking

Running the tests is simple:

crystal spec

If you've saved a brain file with hailo-cr --brain whatever.brain, you can easily inspect it since it's just a SQLite file:

sqlite3 whatever.brain
sqlite> .schema
sqlite> select * from token limit 10;

Support

You can ask a question in the issue tracker, email me at hinrik.sig@gmail.com, or hit me up on FreeNode (#hailo).

Contribute

Pull requests are welcome.

License

hailo.cr is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

hailo.cr is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

See COPYING for the complete license.