Student Enquiry Chatbot source code

The problem it solves

Students often struggle to find accurate information about admission, registration, fees, timetables, results, staff contacts and departmental rules. A local chatbot that works without internet gives instant answers from department-approved content.

Project topic: Development of an AI Chatbot for Answering Student Enquiries in a University Department

Who uses it

What it does

• Ask a question

  Public page where a student types a question. The system cleans the text, matches it against stored answers using keyword scoring, and shows the best answer with its score. Under a matched answer the student can click Yes or No for 'Was this answer helpful?', and the choice is saved. If no score reaches 0.4, it shows an 'I don't know' message with up to three suggested questions. Every question is logged.

• Manage answers and categories

  Staff only. One screen to list, add, edit and delete answers and categories. Each answer has a category, question title, answer text and keywords. The answer form sits beside the answer list, with a second form for categories.

• Enquiry log and most asked

  Staff only. One screen showing every question asked, the matched answer title, score and time, plus a count of the most asked questions and the helpful and not helpful answer counts on the same page.

• Frequently asked questions

  Read-only paginated list of all answers, ordered by category. Lecturers can only open this page.

• Login

  Standard login screen with hashed passwords and CSRF protection. Assigns the correct role to the session.

Screenshots

Screenshots of the running system, taken by our checks. The records are sample data and are labelled so on each page.

Database tables

• usersStores staff and lecturer logins with hashed passwords.id, username, password_hash, full_name, role, created_at
• categoriesOrganises answers into seven fixed topics.id, name, description
• answersOfficial answers written by staff, with keywords used for matching.id, category_id, title, answer_text, keywords, created_at, updated_at
• enquiriesLogs every question typed by a student, with the matched answer and score.id, question, answer_id, score, asked_at, helpful

Technology

• Python. Standard for the course and works fully offline.
• Flask. Lightweight and easy to explain.
• Jinja. Renders HTML pages with variables and loops.
• SQLite. File based, no server, and included with Python.
• Bootstrap. Works offline and looks clear on a laptop and projector.
• Chart.js. Draws simple bar charts without an internet connection.
• Werkzeug. Secure storage of staff logins.
• Flask-WTF CSRFProtect. Prevents cross-site request forgery.

How to run it: running a Python project on Windows. Not sure which language to pick? Python or PHP.

How it was tested

28 of 28 tests passed, 23 of them automated. The system was also installed from scratch and every screen was opened as the right kind of user before it was accepted.

What it does not do

• The matcher compares whole keywords only, so it does not understand meaning, word endings or spelling mistakes. The word 'requirement' does not match the keyword 'requirements'.
• The 0.4 threshold is a demonstration value chosen for this build and is not calibrated.
• A question with no matching keyword gets no evidence at all, so the chatbot says it does not know rather than guessing. This is intended.
• Every answer must be written and maintained by a person. The system never writes or learns an answer by itself.
• The system is a teaching demonstration, not official university advice. All sample answers are fictional.
• Only plain text answers are supported. There are no images and no rich formatting.
• There is no spelling correction, no stemming and no synonyms list, so a student must use some of the stored keywords.
• The log stores the question text only, so nothing links a question to a named student.
• Feedback is offered only when the chatbot matched an answer. A no-match response has no answer to rate, and the helpful column stays null for that enquiry.
• The system does not connect to the internet, does not send email, does not take payments, does not register students for courses and does not replace the official university portal.
• The offline browser checks in check-report.json ran in a Linux sandbox. Windows was not tested here, so a student must confirm run.bat on Windows before the defence.