Smart Scholar- AI Study Assistant

This paper proposes Smart Scholar, a novel integrated AI-powered study assistant that aims to facilitate self-learning through the integration of large language models and retrieval-augmented generation, along with semantic search and a vector-based content store. The system processes a wide variety of study content, such as PDF and slides, and processes them into a vector database, answering student queries through a combination of a retriever and generator. The system’s integrated tools include document summarization, automatic creation of flashcards and questions with auto-evaluation, multimodal explanations, student learning plans, and basic learning analytics for monitoring and adapting to student progress. A prototype and formative user testing show promising results for the system’s potential in enhancing students’ perceived understanding and efficiency of study materials compared to traditional methods like note-taking and chatbots. Smart Scholar aims to alleviate students’ cognitive burden, deliver accurate and context-relevant answers, and assist students in structured learning plans through the use of off-the-shelf LLMs, thereby providing a resource-efficient solution.