Asking Questions

Get the most out of Scrapalot by learning how to ask effective questions and choose the right search mode for your needs.

Choosing the Right Search Mode

Search Modes Explained

RAG Search (Retrieval-Augmented Generation)

Best for: Quick questions about your uploaded documents

How it works:

1. Your question is converted to a vector embedding
2. Scrapalot finds the most relevant document chunks
3. AI generates an answer using only those chunks
4. You get citations showing exactly where the info came from

Example questions:

• "What is the main conclusion of this paper?"
• "Summarize the methodology section"
• "What does the document say about climate change?"
• "List all the key findings"

Typical response time: 2-5 seconds

Settings to adjust:

• Similarity threshold (0.5-0.9): How closely chunks must match your question
• Top K (5-30): Number of document chunks to retrieve
• Strategy: Similarity, HyDE, Multi-Query, etc.

RAG + Web Search

Best for: Questions that combine your documents with current web information

How it works:

1. Searches your documents (like RAG)
2. Also searches the web for additional context
3. Combines both sources in the answer
4. Citations from both documents and web sources

Example questions:

• "How does this paper compare to current industry trends?"
• "What are recent developments related to this topic?"
• "Find statistics to support these claims"

Typical response time: 30-60 seconds

Deep Research

Best for: Complex topics requiring comprehensive research and validation

How it works:

1. Creates a research plan with multiple sections
2. Searches multiple sources in parallel
3. Validates source credibility
4. Synthesizes findings into a structured report
5. Includes quality metrics and citations

Example questions:

• "Provide a comprehensive analysis of renewable energy adoption in Europe"
• "Compare different approaches to AI safety across multiple sources"
• "Research the economic and social impacts of remote work"

Typical response time: 2-5 minutes

See: Deep Research Guide for full details

Chat Mode

Best for: General conversation, brainstorming, or when you don't have relevant documents

How it works:

• Direct conversation with the AI
• No document retrieval
• Uses the AI's general knowledge
• Fast and conversational

Example questions:

• "Explain quantum computing in simple terms"
• "Help me brainstorm ideas for a presentation"
• "What are some good practices for writing documentation?"

Typical response time: 2-5 seconds

Question Types & Techniques

Factual Questions

Ask for specific information from your documents.

Good examples:

• "What is the definition of X in this document?"
• "What year was the study conducted?"
• "Who are the authors of this paper?"

Tips:

• Use specific keywords from your documents
• Be direct and concise
• Use RAG Search mode

Summary Questions

Get overviews and condensed information.

Good examples:

• "Summarize the main points of this document"
• "What are the key findings?"
• "Give me a brief overview of chapter 3"

Tips:

• Specify the level of detail you want
• Indicate if you want bullet points or prose
• Use RAG Search for single documents

Comparative Questions

Compare information across documents or with external sources.

Good examples:

• "Compare the methodologies used in these two papers"
• "How does this approach differ from industry standards?"
• "What are the similarities between documents A and B?"

Tips:

• Select multiple documents or collections
• Use RAG + Web Search for external comparisons
• Be specific about what aspects to compare

Analytical Questions

Request deeper analysis and insights.

Good examples:

• "What are the implications of these findings?"
• "Analyze the strengths and weaknesses of this approach"
• "What patterns emerge from these case studies?"

Tips:

• Use Deep Research for complex analysis
• Increase research depth setting (3-5)
• Allow more time for comprehensive analysis

Research Questions

Complex, multi-faceted questions requiring extensive research.

Good examples:

• "What is the current state of AI safety research?"
• "Analyze the relationship between remote work and urban development"
• "Comprehensive review of climate change mitigation strategies"

Tips:

• Always use Deep Research mode
• Set research depth to 4-5
• Set research breadth to 4-5 for diverse sources
• Expect 3-5 minute response time

Getting Better Answers

Question Flow

Do's and Don'ts

Do:

• ✅ Be specific and clear
• ✅ Include relevant context
• ✅ Specify the format you want (list, paragraph, table)
• ✅ Mention time periods, regions, or specific aspects
• ✅ Use follow-up questions to dig deeper
• ✅ Check citations to verify information

Don't:

• ❌ Ask vague or overly broad questions
• ❌ Expect answers about documents you haven't uploaded
• ❌ Use Deep Research for simple lookups (wastes time)
• ❌ Ignore the suggested mode recommendations
• ❌ Ask multiple unrelated questions in one message

Writing Effective Questions

Instead of:

"Tell me about this"

Try:

"Summarize the main methodology and findings from the 2024 climate report"

---

Instead of:

"What does it say?"

Try:

"What specific recommendations does the document make for small businesses?"

---

Instead of:

"Compare these"

Try:

"Compare the cost-effectiveness and implementation timelines of the three proposed solutions in documents A, B, and C"

Advanced Techniques

Using Collections Effectively

Group related documents into collections for better results:

1. Topic-based collections: All documents about the same subject
2. Time-based collections: Documents from specific time periods
3. Source-based collections: Papers from the same author or organization

Benefits:

• Faster, more focused searches
• Better context understanding
• Easier to manage large document sets

Follow-up Questions

Build on previous answers:

Initial: "What are the main findings of this research?"

Follow-up: "Can you elaborate on the third finding?"

Follow-up: "How does that compare to the Jones 2023 study?"

The AI maintains conversation context, so each follow-up is informed by previous answers.

Adjusting Settings

Fine-tune your queries with these settings:

Similarity Threshold (0.5-0.9):

• 0.5-0.6: Broader search, may include less relevant chunks
• 0.7-0.8: Balanced (recommended for most queries)
• 0.9: Very strict, only highly relevant chunks

Top K (5-30):

• 5-10: Focused, concise answers
• 10-15: Balanced (default)
• 20-30: Comprehensive, longer answers

Research Depth (1-5):

• 1-2: Quick research
• 3: Standard (recommended)
• 4-5: Comprehensive research

Troubleshooting

"No relevant documents found"

Causes:

• Question doesn't match document content
• Similarity threshold too high
• Documents not fully processed

Solutions:

1. Rephrase your question using terms from your documents
2. Lower similarity threshold to 0.6-0.7
3. Check document processing status
4. Verify you've selected the right collection

Answers are too generic

Causes:

• Top K too low
• Question too broad
• Not enough relevant chunks retrieved

Solutions:

1. Increase Top K to 15-20
2. Make your question more specific
3. Lower similarity threshold slightly
4. Select specific documents instead of entire collection

Answers are too long/detailed

Causes:

• Top K too high
• Question asking for comprehensive analysis

Solutions:

1. Reduce Top K to 8-12
2. Ask more focused questions
3. Request bullet points or summaries explicitly

Deep Research taking too long

Expected: Deep Research takes 2-5 minutes

If longer:

• Check your research depth/breadth settings
• Complex topics may take up to 6 minutes
• Monitor progress indicators to see current phase

Related Topics

• Deep Research Architecture - How Deep Research works
• RAG Strategy - Understanding RAG search modes
• Uploading Documents - Prepare your document library
• Features Overview - All Scrapalot features

---

Tip: Start with simple questions to understand your documents, then use Deep Research for complex analysis when needed.