Natural Language Queries: Talk to Your Data

AUM's Natural Language Query (NLQ) engine lets you ask questions in plain English - no SQL required.

How It Works

AUM uses advanced NLP to understand your intent and translate it into data operations.

Supported Query Types

1. Top N Queries

Examples:

• "top 10 products by revenue"
• "show me the best 5 salespersons"
• "highest selling categories"

What Happens:

Intent: RANK
Metric: revenue/sales
Dimension: product/salesperson/category
N: 10/5/default

2. Aggregation Queries

Examples:

• "total revenue"
• "average order value"
• "sum of sales by region"

What Happens:

Intent: AGGREGATE
Operation: sum/average/count
Metric: revenue/order_value/sales
Group By: region (if specified)

3. Group By Queries

Examples:

• "revenue by month"
• "count customers by state"
• "sales by product and region"

What Happens:

Intent: GROUP_BY
Metric: revenue/customers/sales
Dimensions: month/state/[product, region]

4. Filtering Queries

Examples:

• "show orders above 1000"
• "customers in California"
• "products with rating > 4.5"

What Happens:

Intent: FILTER
Condition: amount > 1000, state = 'California', rating > 4.5

5. Trend Analysis

Examples:

• "sales trend over time"
• "show monthly revenue growth"
• "customer acquisition by quarter"

What Happens:

Intent: TREND
Metric: sales/revenue/customers
Time Dimension: auto-detected date column

6. Comparison Queries

Examples:

• "compare sales 2023 vs 2024"
• "revenue this month vs last month"
• "product A vs product B performance"

What Happens:

Intent: COMPARE
Metric: sales/revenue
Segments: 2023 vs 2024, current vs previous

Query Suggestions

AUM automatically generates relevant query suggestions based on your data:

For E-commerce Data:

• "top 10 products by revenue"
• "orders by month"
• "average order value by customer segment"
• "return rate by category"

For Manufacturing Data:

• "OEE by production line"
• "defect rate trend over time"
• "downtime hours by machine"
• "throughput by shift"

For Healthcare Data:

• "bed occupancy by department"
• "average length of stay by diagnosis"
• "appointment no-show rate by clinic"
• "readmission rate trend"

Advanced Features

Column Fuzzy Matching

AUM intelligently matches your query terms to actual column names:

Your Query: "show revenue by customer" Actual Columns: customer_id, total_revenue Matched: ✓ Automatically maps "customer" → "customer_id", "revenue" → "total_revenue"

Multi-Word Column Handling

Your Query: "sales by product category" Actual Column: product_category Matched: ✓ Handles compound words intelligently

Synonym Recognition

• "revenue" = "sales" = "income"
• "customers" = "clients" = "buyers"
• "orders" = "transactions" = "purchases"

Query Response Format

Every query returns:

{
  "success": true,
  "result": {
    "columns": ["product", "revenue"],
    "data": [
      { "product": "Widget A", "revenue": 125000 },
      { "product": "Widget B", "revenue": 98000 }
    ],
    "total_rows": 2,
    "query_type": "rank"
  },
  "insights": [
    {
      "type": "result",
      "insight": "Top 2 products account for 45% of total revenue",
      "importance": 0.8
    }
  ],
  "visualization_suggestions": [
    {
      "type": "bar",
      "x": "product",
      "y": "revenue",
      "description": "Bar chart showing revenue by product"
    }
  ]
}

Using the Query Interface

Desktop

1. Click the search bar at the top of your dashboard
2. Start typing your question
3. See suggested completions
4. Press Enter to execute

Voice Input (Coming Soon)

AUM will support voice queries: "Alexa, ask AUM what my top products are"

Query History

Access your recent queries:

• Click the clock icon in the query bar
• See last 20 queries
• Re-run with one click
• Save favorites

Keyboard Shortcuts

• Cmd/Ctrl + K: Open query bar
• Cmd/Ctrl + Enter: Execute query
• ↑/↓: Navigate query history

Tips for Better Results

✅ DO:

• Use industry-standard terms
• Be specific: "revenue in Q4" > "money"
• Include units: "sales in thousands"
• Specify time ranges: "last month", "2024"

❌ DON'T:

• Use abbreviated slang
• Mix multiple unrelated metrics
• Expect mind-reading (be clear!)

Troubleshooting

Q: My query returned no results A: Check if the column names match your query terms. Try using actual column names.

Q: Wrong columns were selected A: Add more context. Instead of "show sales", try "show total_sales by region"

Q: Can I use complex SQL-like queries? A: Currently, AUM focuses on common business questions. Complex joins and subqueries coming soon!

API Usage

Execute queries programmatically:

const response = await fetch('/api/query/execute', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    'Authorization': 'Bearer YOUR_TOKEN'
  },
  body: JSON.stringify({
    file: fileBlob,
    prompt: 'top 10 products by revenue',
    columns: ['product_name', 'total_revenue']
  })
});

const result = await response.json();

Next: Explore Smart Joins