After building and deploying queries, Infactory’s Explore page allows you to test how they respond to natural language questions. This interface helps you understand query performance, refine your query selection, and create better user experiences.

​
Understanding the Explore Page

The Explore page is your interactive testing ground for deployed queries: The Explore page consists of these main components:
  • Question Input: Where you enter natural language questions
  • Response Panel: Shows the structured data response to your question
  • Query Selection Details: Displays which query was matched to your question
  • Parameter Extraction: Shows how parameters were extracted from your question
  • Alternative Queries: Other queries that might answer your question

​
Testing with Natural Language Questions

The Explore page simulates how end-users will interact with your APIs through natural language:
1

Navigate to the Explore tab

Click on the Explore tab in your Infactory project.
2

Enter a natural language question

Type a question related to your data in the input field.Example: “What is the average height of players by position?”
3

View the response

The system will:
  1. Select the most appropriate query
  2. Extract parameters from your question
  3. Execute the query
  4. Display the structured data results
4

Review query selection details

See which query was selected to answer your question and why.

​
Interpreting Results

Understanding the results helps you refine your queries and improve user experience:

​
Structured Data Response

The primary output is the structured data answer to your question: This data is what would be returned to your application when using the unified API endpoint.

​
Query Selection Information

Below the results, you’ll see details about which query was selected: This includes:
  • The name of the selected query
  • How parameters were extracted from your question
  • The confidence score for this selection
  • Execution metrics (time, data processed)

​
Alternative Queries

Infactory may identify multiple queries that could potentially answer your question: Examining these alternatives can help you:
  • Understand how the system chooses between similar queries
  • Identify opportunities to make your queries more distinct
  • Discover edge cases where the wrong query might be selected

​
How Query Selection Works

Understanding how Infactory selects queries helps you optimize your query library:
1

Question Analysis

Infactory analyzes the natural language question to understand the user’s intent.
2

Query Matching

The system compares the question against your deployed queries’ patterns.
3

Parameter Extraction

For matching queries, Infactory extracts relevant parameters from the question.
4

Confidence Scoring

Each potential match receives a confidence score based on how well it fits.
5

Query Selection

The highest-confidence query that can extract all required parameters is selected.

​
Handling Unanswerable Questions

Not all questions can be answered by your deployed queries. The Explore page helps you identify these gaps: When a question can’t be answered, you’ll see:
  • A clear “No matching query found” message
  • The closest matching queries (if any)
  • Suggestions for how the question might be reworded
This information helps you identify opportunities to create new queries or modify existing ones.

​
Query Coverage Analysis

The Explore page helps you analyze how well your queries cover potential user questions:

​
Testing Different Question Phrasings

Try asking the same question in different ways to ensure your queries are robust:
  • “What’s the average height of quarterbacks?”
  • “Show me the mean height for QB position”
  • “Average QB height”

​
Identifying Coverage Gaps

Systematically test questions about your data to identify areas where you need additional queries: Keep a list of question types that your current queries don’t address, and use it to prioritize new query development.

​
Best Practices for Exploration

Start Broad, Then Refine

Begin with general questions, then get more specific to test query boundaries.

Test Edge Cases

Try unusual phrasings and parameter values to ensure robust handling.

Simulate User Behavior

Ask questions the way your actual users would phrase them.

Document Findings

Keep notes on what works well and what needs improvement.

Test After Changes

Re-test questions after making changes to queries to ensure they still work.

Collaborative Testing

Have team members with different perspectives try asking questions.

​
Fine-Tuning Query Selection

If you find that Infactory isn’t consistently selecting the right query, consider these adjustments:

​
1. Clarify Query Descriptions

Each query has a description that affects matching. Make these descriptions clear and distinct:

​
2. Refine Parameter Names

Clear parameter names improve extraction:

​
3. Add Example Questions

You can add example questions to each query to improve matching:

​
4. Create More Specialized Queries

If queries are too general, they may compete with each other. Create more specialized queries with clearer boundaries:

​
From Exploration to Integration

The Explore page is a powerful tool for iterative development:
  1. Build queries based on your understanding of user needs
  2. Explore how they respond to various questions
  3. Refine your queries based on exploration findings
  4. Deploy the improved queries
  5. Integrate them into your applications with confidence

​
Next Steps

After thoroughly exploring your queries and ensuring they work as expected, you’re ready to integrate them into your applications. Continue to our API Integration Guide to learn how to seamlessly connect your applications with Infactory’s intelligent data APIs.