Due: Wednesday, March 05, 2025

This assignment is worth 100. For full credit:

• submit responses to BrightSpace the end of day on the due date
• see the Deliverables section below

Reading:

• Chatterbot at Github (scroll down past the directory structure)
• Chatterbot Documentation and Quick Start Guide
  

Purpose. To start the semester we looked at ELIZA and other chatbots as they related to the quest to fool interviewers in Turing's Imitation Game. Now we'll look at techniques for building and training a chatbot of your own.

Setup

Follow these steps, which I tested on both Linux and MacOS. I'm using the python3 command to run the python interpreter, since that seems to work most places. If that doesn't work for you, try just python.

1. Do not activate the Anaconda environment. If your prompt in Terminal starts with (base) then run conda deactivate.
2. Change to the directory where your CS356 programs go. Make a new directory called ChatBot, then change to the new directory.
3. Run this command (that ends with a dot, and the spaces are important!):  python3 -m venv . 
4. Assuming it works, list the directory with ls -l to see what new files/directories it created.
5. Now run this command: source bin/activate
6. At this point your prompt should start with (ChatBot). This is a new virtual environment. Now run this: python3 -m pip install chatterbot
7. If everything goes according to plan, the install will add quite a lot of stuff to your virtual environment. Chatterbot can get pretty involved, so it adds almost everything it needs to train chatbots and communicate with different systems, like databases. Let me know if anything fails to work properly, and keep track of error messages.
8. We're still missing one thing: the English language training data. This will download it: python -m spacy download en
9. You should be ready to go! Download this file and give it a try.

Minimum Requirements

To get an idea of how the library works, read through the Quick Start Guide and try out the examples. I did this in two parts:

• simplebot.py
• list_trainer.py

These examples only give one response - not very helpful for training. You can extend each with a loop so you can interact as long as you want. Of course, these are going to be pretty dumb, and the conversation gets dull rapidly.

Now move on to the examples from the Tutorial page. Warning: some will not all run as-is; the installation is missing a few pieces.

The minimum requirements is to produce 5 different chatbots in 5 different Python files: the two above extended with a loop for more interaction, and the results of experimenting with at least 3 different extensions:

• adding at least 10 sentences to those given in list_trainer.py
• experimenting with the ChatterBotCorpusTrainer (see Training and the initial example)
• experimenting with different LogicAdapters

Plan

Remember that in theory, the more you interact with a chatterbot the more it learns how to respond.

Deliverables.

• Post all files needed to build and run each chatbot to the Assignments page on BrightSpace by 4:30pm on the due date
• Include a file README that explains how you trained the chatbot, describes how to run it, and details any bugs or issues you weren't able clear up
• Make sure your name is at the top of each file in a docstring.