kb-mcp-server

You are an expert in information retrieval and question generation. I have provided you with a Markdown document below. Your task is to read and analyze the document carefully, then generate a set of 10 questions based on its content. These questions will be used to test an embeddings/graph-based search system (graph RAG) that processes the document. The system retrieves contexts from the document to answer queries, and I need to evaluate its performance in terms of relevance (returning accurate content), noise reduction (excluding irrelevant sections), precision (delivering focused answers), and completeness (covering key details). ### Instructions: 1. **Analyze the Document**: Identify key topics, concepts, relationships, and specific details within the document. 2. **Generate Diverse Questions**: Create 10 questions that vary in type and complexity, including: - **Factual Questions**: Test retrieval of specific details or definitions (e.g., "What is X?"). - **Relational Questions**: Test understanding of connections between concepts (e.g., "How does X relate to Y?"). - **Explanatory Questions**: Test causal or process explanations (e.g., "Why does X improve Y?" or "How does X work?"). - **Comparative Questions**: Test differentiation of related concepts (e.g., "What’s the difference between X and Y?"). - **Broad Questions**: Test completeness by requiring multiple aspects (e.g., "What are the main benefits of X?"). 3. **Ensure Testability**: Each question should: - Have clear, relevant answers in the document to assess precision and relevance. - Potentially retrieve related but irrelevant content to test noise reduction. - Require specific or connected details to evaluate completeness and graph traversal. 4. **Avoid Ambiguity**: Make questions specific enough to have a single correct focus but not so narrow that they limit retrieval variation. 5. **Reflect Document Structure**: Base questions on headers, sections, and key terms to align with Markdown parsing. ### Output Format: Provide the 10 questions in a numbered list, each prefixed with its type (e.g., "[Factual]", "[Relational]"). After each question, include a brief note (in parentheses) explaining what aspect of the search system it tests. ### Markdown Document: [Insert the full text of your Markdown document here] ### Example (for reference, do not include in output): If the document were about "Machine Learning": 1. [Factual] What is supervised learning? (Tests retrieval of a specific definition) 2. [Relational] How does supervised learning relate to unsupervised learning? (Tests concept connections) 3. [Explanatory] Why does feature scaling improve model performance? (Tests causal reasoning) 4. [Comparative] What’s the difference between linear regression and logistic regression? (Tests precision in distinguishing concepts) 5. [Broad] What are the main applications of neural networks? (Tests completeness across sections) Now, generate the 10 questions based on the provided document.