\n ## Diagram: LLM-based QA Systems - Conceptual Evolution ### Overview The image presents a comparative diagram illustrating the evolution of Question Answering (QA) systems, from conceptual agents to LLM-based agents. It's divided into four sections (A, B, C, and D), each representing a different approach. The diagram uses flowcharts to depict the process and interactions within each system. ### Components/Axes The diagram doesn't have traditional axes. Instead, it uses labeled boxes and arrows to represent components and their relationships. Key components include: * **Agent State (S_t):** Represented in (A). * **Goal:** Represented in (A). * **Naive Agent (A_t):** Represented in (A). * **Action Planning (π):** Represented in (A) and (D). * **Transition Function (T):** Represented in (A). * **Environment:** Present in all sections (A, B, C, and D). * **Question:** Input to QA systems (B, C, and D). * **Answer:** Output from QA systems (C and D). * **LLM QA System:** Overall system in (B). * **Question Understanding Module:** Component of (B). * **Retrieving Module:** Component of (B). * **Answer Generating Module:** Component of (B). * **Information Retrieving:** Input to (B). * **Tool Using:** Input to (B). * **Retrieved Information:** Output of (B). * **Tool Results:** Output of (B). * **LLM:** Core component of (C). * **LLM QA Agent:** Core component of (D).