This chatbot will initiate a conversation with the user, interacting with them like a friend. It will be able to understand the gist of what the user is saying, and (hopefully) respond in an appropriate way.

The simplest way is to run python -m pip install nltk Then, a few NLTK modules must be downloaded. They can be downloaded by running Python in the command line and typing the following:

import nltk
nltk.download('punkt')
nltk.download('wordnet')
nltk.download('vader_lexicon')
nltk.download('averaged_perceptron_tagger')

All conversation topics are handled dynamically in convo.dat. The first line of a topic is structured as follows:

. or ? keywords (^variable.wordType OR +)

. or ?: Depending on whether the topic expects a statement (.) or question (?) from the user (although currently unused).

keywords: When looking for only one of a multiple set of similar keywords, separate them with /'s (eg, hi/hey/hello). When looking for another set/single keyword, separate with &'s (eg, i&like&ham). Note that they must all be lowercase. Combined example: do&you/u&like/enjoy&soccer/football

^variable: Used when expecting to detect a single word from the user to save in memory, referenced by the give name ('variable' in this case)

.wordType: Dictates which word type (as defined by the NLTK library. eg, NNP = proper noun, VB = verb) to look for in the user's response, in order to help determine which word to save as the answer 'variable'.

+: Used when expecting to detect the positivity of the user's following response. Note: Only one of ^variable.wordType and + may be used at a time. In other words, the chatbot cannot handle both determining the positivity of the user's next response as well as saving a certain answer from it.

The rest of the topic is separated by newlines, with the bot waiting a response between each line before printing the next line.

In order to reference a saved answer from the user (^variable from before), simply put $variable in the bot's response, which will be replaced by the saved answer. Note that an answer must first be saved before it can be called like this. Otherwise, it will return "NaN" instead.

When using the + operator, the following 3 lines must start with + (positive), - (negative), and 0 (neutral), in any order. The positivity score of the user's last response will be used to determine which of these three responses to output.

convo.dat file must begin with an empty line, and end with two empty lines.

First 2 topics of convo.dat must be default statement response and default question response, respectively.

Correlation algorithm likely won't scale well when many more topics are added.

Doesn't handle context well; user must explicitly mention what they're referring to each time they ask a question.

Can't both save an answer and determine the positivity of a single user response.

If the + operator is used, the next 3 lines MUST start with each one of +, -, and 0.

1. Take context into account

2. Create better correlation system

3. Make bot able to give some suggested topics/questions for the user to ask.

4. Make certain words weighted heavier (why, what, etc) when choosing topics

5. Use first character of topics (. or ?) to weight responses

6. Create and link non-CLI UI

7. Allow wildcards in topic keywords (eg, 'fav*' matches fave, favorite, and favourite)