Chatbot-QueryBot-Python

This repository contains multiple versions of a chatbot built using Natural Language Processing (NLP) techniques. The chatbot processes input from users and predicts appropriate responses based on predefined intents. The models utilize different machine learning frameworks, including tflearn and Keras.

Versions:

Version 1: chatbot-QueryBot-V1

• Framework: tflearn
• Model Type: Neural Network
• Training Data: Uses intents data stored in the intents.json file.

Files:

• main.py: The main script to train and run the chatbot.
  • Downloads necessary NLP packages (nltk).
  • Preprocesses the input text and prepares the training data.
  • Uses tflearn to build a neural network with two hidden layers.
  • Saves the trained model to model.tflearn file.

Version 2: chatbot-QueryBot-V2

• Framework: Keras and tflearn
• Model Type: Neural Network (with both Keras and tflearn implementations)
• Training Data: Uses intents data stored in the intents.json file.

Files:

• ChatBot_NLP_Model_keras.py: A Keras-based version of the chatbot.

  • Similar structure to Version 1, but uses Keras layers to build the model.
  • Trains the model for a larger number of epochs and saves it as model.h5.

• ChatBot_NLP_Model_tflearn.py: An alternative tflearn implementation.

  • Uses tflearn to create a model with multiple hidden layers.
  • Trains the model and saves it as model.tflearn.

Version 3: chatbot-QueryBot-V3 (New)

• Framework: Under Development
• Model Type: Improved version with enhanced features.
• Training Data: Uses intents data stored in intents.json and additional datasets.
• Note: DataGenerator_Beta.py is still under development and is not fully functional at this stage.

Files:

• chatbot-QueryBot-V3.py: The main script for the latest version of the chatbot, using updated techniques.
• DataGenerator_Beta.py: Currently under development; not yet fully operational.
• Chatbot_Model_Fuzzy_Matching.py: Implements fuzzy matching for more flexible response generation.
• Chatbot_Model_Intent_Classifier.py: An updated classifier to improve intent detection.
• DataGenerator.py: A utility for generating and processing data for the chatbot.
• intents.json: JSON file containing predefined intents and responses.
• dataset.json: A dataset to enhance training with additional examples.

Requirements:

• Python 3.x
• Install necessary libraries:
  • tensorflow
  • tflearn
  • nltk
  • numpy
  • json
  • pickle

You can install the required dependencies with the following:

pip install tensorflow tflearn nltk numpy

Running the Chatbot:

1. Download or clone the repository.
2. Navigate to the folder containing the project files.
3. Run any version:

• For Version 1:

python chatbot-QueryBot-V1/main.py

• For Version 2 (Keras implementation):

python chatbot-QueryBot-V2/ChatBot_NLP_Model_keras.py

• For Version 2 (tflearn implementation):

python chatbot-QueryBot-V2/ChatBot_NLP_Model_tflearn.py

• For Version 3 (if available):

python chatbot-QueryBot-V3/chatbot-QueryBot-V3.py

4. Enter text and interact with the bot. Type "quit" to exit the conversation.

Model Training:

• When you run the chatbot for the first time, it will train the model using the data from intents.json and save it as data.pickle.
• If the model is pre-trained and the pickle file exists, it will directly load the model data.

Usage:

• The chatbot responds to various patterns defined in the intents.json file. Each pattern is associated with a tag, and the chatbot predicts the most likely response based on the user’s input.

Intents Format:

The intents.json file defines different patterns and their corresponding responses. An example format:

{
    "intents": [
        {
            "tag": "greeting",
            "patterns": ["Hi", "Hello", "How are you?"],
            "responses": ["Hello!", "Hi, how can I help?", "Greetings!"]
        },
        {
            "tag": "goodbye",
            "patterns": ["Bye", "See you", "Goodbye"],
            "responses": ["Goodbye!", "See you later!", "Take care!"]
        }
    ]
}

Model Architecture:

• Input Layer: Accepts a sequence of words tokenized from the user input.
• Hidden Layers: Two or more fully connected layers with ReLU activation.
• Output Layer: A softmax layer that outputs probabilities for each intent.

Future Improvements:

• Adding more advanced NLP models.
• Integrating with external APIs for dynamic responses.
• Expanding the intents dataset for a broader range of interactions.
• Completing the development of DataGenerator_Beta.py for better data processing.

License:

This project is licensed under the MIT License - see the LICENSE file for details.