NLP Chatbot

A simple NLP-based chatbot application that supports both English and Indonesian languages. The chatbot can perform basic greetings, answer questions about its creator, and handle arithmetic operations. The project is organized into modular files to manage training, API, and processing.

Project Structure

• index.js: Main entry point of the application. Handles command-line arguments and initiates different modes.
• train.js: Trains and saves the NLP model.
• model.js: Defines intents and answers for greetings, math operations, and creator information.
• process.js: Processes questions and performs calculations.
• server.js: Sets up an Express server to handle HTTP API requests.
• load.js: Loads the saved NLP model from a file.

Getting Started

Prerequisites

• Node.js installed on your machine.
• npm for package management.

Installation

1. Clone the repository:

   git clone https://github.com/Cloud-Dark/nlpjs

2. Navigate to the project directory:

   cd nlp-chatbot

3. Install the dependencies:

   npm install

Usage

The project supports three main commands: train, api, and simple_qna.

1. Train the Model

Train the NLP model with predefined intents and answers, then save it to a file (brain.model):

node index.js train

2. Start the API Server

After training the model, you can start an API server that listens for HTTP POST requests and responds with answers:

node index.js api

API Endpoint:

• URL: http://localhost:3000/api/message
• Method: POST
• Request Body:

  {
    "language": "en",   // Language code ('en' for English, 'id' for Indonesian)
    "text": "hello"     // The user input text
  }

• Response:

  {
    "answer": "Hey there!"
  }

3. Run a Simple Q&A Command

Use simple_qna to process a single question from the command line:

node index.js simple_qna "<text>" [language]

• <text>: The input text/question.
• [language] (optional): Language code, either en (default) or id.

Example:

node index.js simple_qna "hi" en
node index.js simple_qna "siapa pembuat mu" id

Project Modules

Each module serves a distinct purpose, keeping the project organized and modular:

• train.js: Initializes, trains, and saves the model.
• model.js: Contains all predefined intents and responses.
• process.js: Processes user input and handles arithmetic operations.
• server.js: Runs an Express server to manage API requests.
• load.js: Loads the saved model (brain.model) for use.

Example Usage

Training the Model

node index.js train

API Example (using curl)

curl -X POST http://localhost:3000/api/message -H "Content-Type: application/json" -d '{"language": "id", "text": "siapa pembuat mu"}'

Simple Q&A Example

node index.js simple_qna "calculate 10 plus 5" en
node index.js simple_qna "hitung 10 ditambah 5" id

License

This project is licensed under the MIT License.