Smart_Chatbot

A lightweight Rule-Based Self-Learning Chatbot built with Python that learns new responses during conversations and stores them locally in a JSON database.

Unlike traditional hardcoded chatbots, this chatbot can remember new question-response pairs provided by the user. It uses fuzzy string matching to recognize similar questions and retrieve the most appropriate response.

Features

• Learn new responses during conversations
• Store learned knowledge permanently in JSON
• Fuzzy matching using Python's difflib.SequenceMatcher
• Multiple responses for the same question
• Random response selection for natural conversations
• Display all learned knowledge
• Simple command-line interface
• No external libraries required

---

How It Works

1. User asks a question.
2. The chatbot searches its knowledge base.
3. If a similar question (90% similarity or higher) exists:
   • A stored response is returned.
4. If no match is found:
   • The chatbot asks the user for the correct response.
   • The new question-response pair is saved into responses.json.
5. Future conversations can reuse the learned response.

---

Project Structure

project/
│
├── main.py          # Main chatbot application
├── responses.json      # Knowledge base
└── README.md

---

Technologies Used

• Python 3
• JSON
• difflib.SequenceMatcher
• Random module

---

Commands

Command	Description
!exit	Exit the chatbot
!bye	Exit the chatbot
!show responses	Display all learned responses
show what you learned	Display the knowledge base

---

Example

You: Hello
Chatbot: I don't know the answer. Can you tell me?

how to respond to 'Hello'
Type Answer here:
Hi! Nice to meet you.

Chatbot: Got it! I'll remember that for next time.

Later...

You: Hello
Chatbot: Hi! Nice to meet you.

---

Learning Process

User Input
     │
     ▼
Search JSON Knowledge Base
     │
     ├──────── Match Found ───────► Return Response
     │
     ▼
 No Match
     │
Ask User for Response
     │
Store in JSON
     │
Future Conversations Use It

---

Current Limitations

• Uses character similarity instead of semantic understanding.
• Cannot understand context.
• No intent classification.
• No natural language processing (NLP).
• Case-sensitive similarity may affect matching.
• Responses depend entirely on previously learned data.

---

Future Improvements

• GUI
• SQLite database

---

Author

Abhishek Kumar