The objective was to test a RAG system's ability to provide concise answers through prompt engineering and parameter optimization. Using four influential works of Immanuel Kant (a challenging domain), I assessed the system's contextual awareness and retrieval capabilities. The tests involved two sets of questions: direct definition queries and complex analysis questions requiring long-span context understanding.
The texts used for the analysis are (obtained from the Project Gutenberg data repository):
Book_1: Kant's Critique of Judgement
Book_2: The Critique of Practical Reason
Book_3: The Critique of Pure Reason
Book_4: Fundamental Principles of the Metaphysic of Morals
For more information, visit my Portfolio: https://jelinr.github.io/