LECTRALLM-Enhanced Course Teaching and Retrieval Assistant
LECTRA is a planned interactive learning platform where students keep course material visible while working with a context-aware LLM assistant. It is designed to provide grounded answers, explanations, examples, practice questions, feedback, and a visible adaptive next step.
Illustration of the planned demonstrator
The question
More useful help than retrieval alone?
Does a lightweight student learning-state model enable more useful and appropriately difficult explanations and exercises than course-material retrieval alone?
LECTRA will model topic-level signals from prior interactions, questions, and performance. These signals are estimates—not grades—and should be transparent enough for students or teachers to understand why practice on a topic was recommended.
The working screen
Keep the material in view while asking
Course material
Slides, PDFs, and a rendered Jupyter notebook stay visible while the student works.
Material-aware conversation
A student selects the relevant slide, page, section, or notebook cell before asking a question.
Grounded help
Answers should point back to the relevant source material, then offer a deeper or simpler explanation and additional examples.
A learning loop, not just a chat
From a question to a targeted next step
1
Ask or practise
Ask about the selected material, request an explanation or example, or start a practice question.
2
Receive feedback
The demonstrator should check an answer, explain mistakes, identify missing concepts, and suggest what to review.
3
Adapt the next step
A lightweight topic-level estimate of understanding, uncertainty, or gaps should shape at least one visible recommendation.
Apply what you learn
Practise with questions and code in context
Students will be able to simulate exam-style questions based on the selected course content, answer them, and receive targeted feedback. When a concept is best understood through its effect on an implementation, the agent should also be able to write code into the Jupyter notebook so that the student can see the explanation alongside its impact on code.
Exam-style practice and notebook code
What the demonstrator must cover
One prepared pilot course workspace
Slide/PDF and rendered notebook viewers
Explicit material context and referenced answers
Explanations, examples, questions, and answer feedback
Topic-level learning-state tracking and one adaptive behaviour
A deployed demonstrator
Final demonstration
A student opens lecture material and a related notebook, follows references from the chat, requests a simpler explanation and examples, answers a generated question, receives targeted feedback, and sees an adaptive follow-up recommendation.
The project is about taking a research question through to a working product demonstrator: connecting course material, interaction design, grounded LLM support, feedback, and a transparent topic-level learning-state estimate.
You are comfortable writing Python.
You are comfortable with git and working in a shared repository.
You enjoy working in a mixed international team across two universities.
Interest in machine learning helps—especially personalisation and learning from user feedback.
Curiosity about how people learn, become uncertain, and benefit from feedback matters as much as model experience.
What you walk away with
The experience of carrying a genuine research question all the way to a working system, in a mixed team across two universities and one shared repository.
