In today’s fast-paced digital environment, individuals frequently experience stress, demotivation, and emotional imbalance due to academic, professional, and personal challenges. Although motivational quotes are widely available online, users must manually search through multiple websites or platforms to find content that matches their emotional state.
This manual process is:
- Time-consuming
- Non-personalized
- Inefficient
- Lacking conversational interaction
There is a need for an intelligent system that can understand user intent or emotion and recommend relevant quotes instantly through natural conversation.
The proposed system should:
- Accept user input in natural language
- Classify user intent using NLP
- Recommend quotes based on identified emotion or category
- Provide interactive conversational responses
- Allow scalability for adding new quote categories
Existing quote platforms such as BrainyQuote and Goodreads provide categorized quotes but rely on manual browsing and keyword search. These systems do not offer real-time personalization or conversational interaction.
Conversational AI frameworks like Rasa enable intent classification and dialogue management using machine learning models such as DIETClassifier. Most chatbot implementations focus on customer service or FAQ automation rather than emotion-based quote recommendation.
There is a clear gap in systems that combine conversational AI with personalized motivational content delivery.
This chatbot can:
- Improve emotional well-being through personalized motivational support
- Increase user engagement through interactive conversation
- Be extended for educational, wellness, or mental health applications
- Serve as a foundation for AI-powered recommendation systems
Before running the Quotes Recommendation Chatbot, ensure the following requirements are installed and properly configured.
- Python 3.8 or above
- Virtual environment recommended
Create virtual environment:
python -m venv venvActivate the environment:
# Windows
venv\Scripts\activate
# macOS/Linux
source venv/bin/activateAnaconda can be used for easier environment and dependency management.
Create a conda environment:
conda create -n chatbot_env python=3.8
conda activate chatbot_envInstall Rasa using pip:
pip install rasaVerify installation:
rasa --versionThe chatbot requires a structured dataset containing:
- Quote text
- Corresponding emotion or category label
Example format (CSV):
| quote | category |
|---|---|
| Never give up | motivation |
| Love conquers all | love |
| This too shall pass | hope |
Training data must follow the Rasa YAML structure.
Example (nlu.yml):
version: "3.1"
nlu:
- intent: sadness
examples: |
- I feel lonely
- I am sad
- I feel low
- intent: motivation
examples: |
- Inspire me
- I need motivation
- Encourage meThe NLP pipeline is defined inside the config.yml file.
Example configuration:
pipeline:
- name: WhitespaceTokenizer
- name: CountVectorsFeaturizer
- name: DIETClassifier
- name: EntitySynonymMapper- WhitespaceTokenizer → Splits user input into tokens
- CountVectorsFeaturizer → Converts text into numerical features
- DIETClassifier → Performs intent classification
- EntitySynonymMapper → Maps similar entities to a standard value
The following tools are recommended for development:
- Visual Studio Code (or any preferred IDE)
- Git for version control
- Basic understanding of Python and NLP concepts
The development of the Quotes Recommendation Chatbot follows a structured milestone-based workflow:
- Specify the business problem
- Define business requirements
- Conduct literature survey
- Analyze social or business impact
- Install Rasa and required dependencies
- Set up the Rasa project structure
- Create user queries (
nlu.yml) - Create bot responses (
domain.yml) - Define conversation stories (
stories.yml) - Train the chatbot model
- Store and manage trained models
- Test model using Rasa shell
- Test using test stories
- Deploy chatbot using web interface
- Validate deployed chatbot
The project follows the standard Rasa directory structure:
quotes-recommendation-chatbot/
│
├── data/
│ ├── nlu.yml
│ └── stories.yml
│
├── domain.yml
├── config.yml
├── credentials.yml
├── endpoints.yml
│
├── models/
│
└── README.md
- data/ → Contains NLU training data and conversation stories
- nlu.yml → User intents and training examples
- stories.yml → Conversation flow definitions
- domain.yml → Intents, entities, responses, and actions
- config.yml → NLU pipeline configuration
- models/ → Stores trained chatbot models
- Emotion-based quote recommendation
- Intent recognition using Rasa NLU
- Structured conversation flow using stories
- Custom NLU pipeline configuration
- Scalable and modular chatbot architecture
- Easy deployment using Rasa shell or web interface
git clone https://github.com/your-username/quotes-recommendation-chatbot.git
cd quotes-recommendation-chatbotsource venv/bin/activate # macOS/Linux
venv\Scripts\activate # Windowsrasa trainrasa shellAfter training the chatbot model, testing was performed using Rasa’s built-in testing framework to evaluate the chatbot’s performance in predicting correct actions and responses.
The chatbot was tested using the following command:
rasa test coreThe evaluation results obtained were:
| Metric | Value |
|---|---|
| Correct Predictions | 58 / 58 |
| Accuracy | 1.000 |
| Precision | 1.000 |
| F1 Score | 1.000 |
These results indicate that the chatbot correctly predicts the appropriate responses for all test stories.
The confusion matrix generated during testing shows that all predicted actions match the expected actions, indicating no misclassification during testing.
- The chatbot correctly predicts actions for all test stories.
- The model achieved 100% accuracy during testing.
- High precision and F1 score indicate correct intent recognition.
- The chatbot successfully returns the appropriate quote category based on user input.
The Quotes Recommendation Chatbot demonstrates how Natural Language Processing and Rasa can be used to build an intelligent, emotion-aware conversational system.
The project follows a structured development lifecycle including problem understanding, environment setup, model training, testing, and deployment.
This chatbot can be extended further by:
- Integrating a web frontend
- Connecting to a database of quotes
- Adding sentiment analysis
- Deploying on cloud platforms