CSCI 150 - Foundations of Computer Science

Fall 2013

Project 3: Chatterbot

Chatterbots

A chatterbot is a program that simulates a conversation with a human user. It relies upon vague statements to create an illusion of an intelligent conversation. It looks for patterns and critical words in user inputs to formulate plausible-sounding responses.

The earliest chatterbot program was called ELIZA. It simulated a Rogerian psychotherapist. Here is an example of a conversation between the instructor and his own simple implementation of ELIZA:

>>> chat(10)
I am Eliza.  How can I help you?
> I am tired.
Why do you feel tired?
> It is late at night.
Tell me more about tired.
> I get tired when it is late at night after a lot of work.
Why do you feel tired?
> Why do you keep asking?
I don't have answers; I give you questions to help you find your own answers.
> I'm sorry.
Tell me more about tired.
> I have too much to do.
Tell me more.
> It is satisfying to finish lots of work.
Why do you find that to be satisfying?
> I feel like I can make my father proud.
Tell me more about your relationship with your father.
> He always pushed me to do my best.
Tell me more about satisfying.
> It is very satisfying to do my best work.
Why do you find that to be satisfying?
> It is a feeling of fulfillment.
Thank you for chatting.  Have a nice day.
>>>

The Project

The skeleton file for the project contains the following:

greeting = "I am Eliza.  How can I help you?"

def chat(n):
    print(greeting)
    for i in range(n):
        print(reply(input("> ")))
    input("> ")
    print("So long for now.  Have a nice day.")

def reply(sentence):
    return "Reply"

Your task is to rewrite the reply() function to have a distinct personality. Each call to reply() will generate one answer to one user sentence. Also, modify greeting to be consistent with your program's character. Here are the requirements:

There must be at least six different types of response.
The specific type of response selected must depend in some way on the user input.
At least three of the responses will incorporate at least one word from a user sentence.
Note that the sample conversation above exhibits all of these criteria.
Lists are very convenient for storing a category of terms for which to search. For example, my implementation has a list of family members. All family members trigger exactly the same response, except that the specific family member named is incorporated into the response.
String methods such as split() and replace() will likely prove to be very useful for this project.
When your project is complete, submit your edited verison of chat.py, including your customized greeting and customized reply() function.
The project is due on Thursday, October 3, 2013 at 11:55 pm.