kunal3 / markov-bot

Created a simple sloppy markov bot when I was bored.

Kunal Agrawal

---

An explanation of the markov depth :

It defines how many words to group before associating them with the following word.

For example, inputString = "My name is Kunal."

markovDepth = 1 Resulting Markov associations - {'is': ['Kunal.'], 'My': ['name'], 'name': ['is']}

markovDepth = 2 Resulting Markov associations - {'My name': ['is'], 'name is': ['Kunal.']}

---

TODO -

• Add cmd line args
• Make it listen to everything everyone says, except the things it itself says.
• Break sentence on punctuations as well. Probably can just split on "" ( ) /". But don't separate period because it will be nice to have knowledge of ends of sentences.
• Add a 5% or something chance to respond to every message. random.randint(1,100)
• Make smart responses by building a backwards markov chains.
• Add a chance to respond with multiple sentences.
• Add some sort of natural language processing for sensible replies
• Link to authorize my bot - https://discordapp.com/oauth2/authorize?&client_id=241600786470010881&scope=bot

---

(WORK IN PROGRESS)

How we plan to have the markov brain talk about a subject :

We have a reverse dictionary for our forward markov chain dictionary. So the above 2 depth association's reverse dictionary would look something like this - {'is': ['My name'], 'Kunal.': ['name is']}

Now, the key is followed by words which precede it. Now we can build a sentence forward and backwards from our two dictionaries!

This most efficient way of searching for the subject word is to have a third dictionary, which takes the keys from the first dictionary and splits it up into single word and associates every pair, just like you would do to the first dictionary if our markov depth was 1.

Building the sentence forward - Take dict 3, find our subject keyword in it, and follow this keyword's values down the rabbit hole markov depth times. Once we have a string of markov depth length, use this as a key for the first dictionary and building your forward sentence.

Building the sentence backwards - Since the backwards dictionary will always have keys with a depth of 1,

markov-bot's Issues

j n