Discord chatbot

Discord chatbot created for the NLP elective course at Insper

How to run it:

Warning: Python version is 3.10

1. First of all, create a discord bot on the discord developer portal. You can follow this tutorial to do it

2. pip install -r requirements.txt

3. then, create a .env file with:

        TOKEN = Bot token
        KEY_RESET = Reset key for the command !reset
        FILE_CLASS_PICKLE = Name of a pickle file like: class1.pkl
        GUILD = Server Name where the bot is
        CHANNEL = Channel that the bot will use
        API_KEY = OpenAI API key
   

4. Download Wordnet using the following command on the python terminal:

        import nltk
        nltk.download('wordnet')
   

5. Now, you are able to run it!

        python3 chatbot.py
   

Chatbot Commands

On ready:

!help:

Show help message describing all commands

!source:

Returns the source code from this bot

!author:

Returns the author's name

!run:

get informations from IP address.

IPv4 format: x.x.x.x

IPv6 format y:y:y:y:y:y:y:y (also works with abreviation.e.g->2001:db8::)

!run [IPv4]

!run [IPv6] v6

API used: https://rapidapi.com/xakageminato/api/ip-geolocation-ipwhois-io

!crawl

web crawling from webpage. Only receive 1 url and craw over a max of 15 pages. There is a timout with request takes more than 15 seconds.

!crawl

!search

Seach for a word or a phrase in the documents and show all the documents where the word appears and the TFiDF (measures how important a term is within a document relative to a collection of documents). You can use the argument th=X in the end for filtering pages by sentiment value. th default is 0 which means that filter wasn't applied. Goes from 0 (negative) to 1 (positive).

!search [word] [th=[value]]

!wn_search

Search for one word in the documents and show all the documents where the word appears and the TFiDF (measures how important a term is within a document relative to a collection of documents). If the word is not in the documents, try to find synonims that is in the documents. You can use the argument th=X in the end for filtering pages by sentiment value. th default is 0 which means that filter wasn't applied. Goes from 0 (negative text) to 1 (positive text).

!wn_search [word] [th=[value]]

!generate

Generate text from database using model

!generate [word]

!get_all_pages

Get the url and sentiment value from all pages that have been webscrapped. You can use the argument th=X in the end for filtering pages by sentiment value. th default is 0 which means that filter wasn't applied. Goes from 0 (negative text) to 1 (positive text).

!reset

Web crawling reset

!reset [secret_key]

Essays:

• Essay 0

• Essay 1

• Essay 2

• Essay 3

• Essay 4

• Essay 5