Mental Health Chatbot

This project focuses on building a mental health chatbot that can provide empathetic responses and offer basic mental health information. The chatbot uses a logistic regression classifier to categorize user queries into two groups: information-seeking (class A) and emotional support-seeking (class B). The chatbot then generates appropriate canned responses based on these classifications.

Table of Contents

• Requirements
• Installation
  • Using Conda
  • Using venv
• Running the Chatbot
• Project Structure
• Schema Definition
• Usage
• Credits

Requirements

• Python 3.8 or higher
• Libraries:
  • absl-py
  • pandas
  • sentence-transformers
  • scikit-learn

Installation

Using Conda

1. Install Conda: Make sure Conda is installed on your machine. If not, download and install it from Conda's website.

2. Create a Conda environment:

conda create -n chatbot_env python=3.8

3. Activate the environment::

conda activate chatbot_env

4. Install dependencies: You can install the required dependencies using the provided requirements.txt file:

pip install -r requirements.txt

5. Installing sentence_transformers Using Conda

conda install -c conda-forge sentence-transformers

if this fails try from anaconda.org:

conda install conda-forge::sentence-transformers

Using venv

1. Install venv (if needed): If you don’t have venv installed, you can do so with the following command:

sudo apt install python3-venv   # On Ubuntu/Debian

2. Create a virtual environment:

python3 -m venv chatbot_env

3. Activate the environment:

   On Linux/macOS:

   source chatbot_env/bin/activate

   On Windows:

   chatbot_env\Scripts\activate

4. Install dependencies:

pip install -r requirements.txt

Running the Chatbot

Once you have installed the necessary dependencies and activated your environment, you can run the chatbot using the following command:

python conversation_chatbot_main.py --faq_data_path ./data/Mental_Health_FAQ.csv --empathetic_data_path ./data/empathetic-dialogues-contexts --conversations_data_path ./data/mental_health_counseling_conversations.csv

alt view:

python conversation_chatbot_main.py --faq_data_path ./data/Mental_Health_FAQ.csv \
               --empathetic_data_path ./data/empathetic-dialogues-contexts/ \
               --conversations_data_path ./data/mental_health_counseling_conversations.csv

Optional Arguments

• --faq_data_path: Path to the FAQ dataset CSV file (default is ./data/Mental_Health_FAQ.csv).
• --empathetic_data_path: Path to the directory containing the empathetic dialogues datasets (default is ./data/empathetic-dialogues-contexts/).
• --conversations_data_path: Path to the CSV file containing mental health counseling conversations (default is ./data/mental_health_counseling_conversations.csv).
• --test_size: Fraction of data to use for testing (default is 0.3).
• --random_state: Seed for random number generation (default is 42).

The chatbot will start running, and you can interact with it via the command line. Type your query, and the chatbot will categorize it as either informational or emotional, providing a suitable response. To exit the chatbot, you can type any of the following:

• exit
• quit
• q
• x
• e
• Or press Ctrl+C (keyboard interrupt)

Project Structure

The project follows a structured organization defined using GraphQL schema. Below is both the visual directory structure and its corresponding schema representation:

mental_health_chatbot/
│
├── main.py                  # Entry point of the application
├── chatbot.py               # Contains the MentalHealthChatbot class
├── utils.py                 # Utility functions
├── data/
│   ├── Mental_Health_FAQ.csv
│   ├── canned_responses.csv    
│   └── empathetic-dialogues-contexts/
│       ├── train.csv
│       ├── valid.csv
│       └── test.csv
├── requirements.txt         # List of dependencies
└── README.md               # Project documentation

Schema Definition

The project structure is formally defined in project-structure.graphql:

# Import from ./project-structure.graphql
type Project {
  name: String!
  rootDirectory: Directory!
}

# See complete schema in project-structure.graphql

Key Components

• Source Files:

  • main.py: Entry point for the chatbot application
  • chatbot.py: Core chatbot implementation
  • utils.py: Helper functions and utilities

• Data Directory:

  • Mental health FAQ dataset
  • Canned responses
  • Empathetic dialogues training data

• Configuration:

  • requirements.txt: Project dependencies
  • README.md: Documentation and setup instructions

Key Files and Directories

• data/: Contains the required datasets for running the chatbot, including FAQ data, counseling conversation data, and empathetic dialogues.
• mental_health_chatbot.py: Implements the MentalHealthChatbot class, handling data loading, preprocessing, model training, and generating responses.
• conversation_chatbot_main.py: Contains the runtime logic and manages the interactive loop for user queries.
• README.md: Provides an overview of the project, installation steps, running instructions, and usage examples.
• requirements.txt: Lists the dependencies needed to run the project (e.g., sentence-transformers, scikit-learn).

Ensure the datasets are correctly placed in the data/ directory as per the structure, and you're ready to run the chatbot using the provided instructions.

Usage

Once the chatbot is running, type any mental health-related query. The chatbot will categorize your query as either informational or emotional. It will then respond with an appropriate answer based on its classification, using the FAQ dataset for informational queries and a conversation graph derived from counseling sessions for emotional queries.

Example interaction:

You: I'm feeling really stressed out lately.
Chatbot: I'm sorry to hear that you're feeling stressed. Would you like to talk about it?

You: What are the symptoms of anxiety?
Chatbot: Some common symptoms of anxiety include nervousness, restlessness, and a sense of impending danger.

Credits

• Sentence-BERT: Used for encoding questions and contexts.
• Scikit-learn: Used for implementing classifiers.
• Pandas: For handling data loading and preprocessing.