Skip to content

Intent Handling

Intent handling detects what users are looking for in their queries (brand, category, attribute, use-case) and applies appropriate ranking behavior for each intent type.

Overview

Different query types require different ranking strategies. Intent handling automatically detects query intent and applies the most appropriate ranking behavior, ensuring users find what they're looking for.

Intent Types

Brand Intent

Users searching for a specific brand:

  • Queries: "Nike shoes", "Apple iPhone", "Samsung TV"
  • Strategy: Prioritize exact brand matches, show brand-specific products
  • Ranking: High weight on brand match, lower weight on other attributes
{
  "intent_type": "brand",
  "detection_patterns": [
    "brand_name + product_type",
    "brand_name only"
  ],
  "ranking_strategy": {
    "brand_match_weight": 0.8,
    "relevance_weight": 0.2
  }
}

Category Intent

Users browsing a category:

  • Queries: "running shoes", "laptops", "kitchen appliances"
  • Strategy: Show diverse products from the category
  • Ranking: High weight on category match, diversity across brands/attributes
{
  "intent_type": "category",
  "detection_patterns": [
    "category_name",
    "product_type"
  ],
  "ranking_strategy": {
    "category_match_weight": 0.7,
    "diversity_weight": 0.3,
    "apply_guardrails": true
  }
}

Attribute Intent

Users searching by specific attributes:

  • Queries: "red dress", "wireless headphones", "waterproof watch"
  • Strategy: Prioritize products matching the attribute
  • Ranking: High weight on attribute match, filter by attribute
{
  "intent_type": "attribute",
  "detection_patterns": [
    "color + product_type",
    "material + product_type",
    "feature + product_type"
  ],
  "ranking_strategy": {
    "attribute_match_weight": 0.9,
    "relevance_weight": 0.1
  }
}

Use-Case Intent

Users searching for a use case or problem:

  • Queries: "gift for mom", "workout gear", "home office setup"
  • Strategy: Show products that solve the use case
  • Ranking: Semantic relevance + use-case tags
{
  "intent_type": "use_case",
  "detection_patterns": [
    "for + use_case",
    "gift for",
    "best for"
  ],
  "ranking_strategy": {
    "semantic_relevance_weight": 0.6,
    "use_case_tags_weight": 0.4
  }
}

Comparison Intent

Users comparing products:

  • Queries: "iPhone vs Samsung", "best running shoes"
  • Strategy: Show top products for comparison
  • Ranking: High-quality, comparable products
{
  "intent_type": "comparison",
  "detection_patterns": [
    "vs",
    "compare",
    "best"
  ],
  "ranking_strategy": {
    "quality_weight": 0.5,
    "popularity_weight": 0.3,
    "reviews_weight": 0.2
  }
}

Intent Detection

Detect intent using multiple methods:

Pattern Matching

{
  "detection_method": "pattern_matching",
  "patterns": {
    "brand": ["brand_name + product", "brand_name"],
    "category": ["product_type", "category_name"],
    "attribute": ["attribute + product", "color + product"]
  }
}

Machine Learning Classification

{
  "detection_method": "ml_classification",
  "model": "intent_classifier_v1",
  "features": [
    "query_text",
    "query_length",
    "contains_brand",
    "contains_category",
    "contains_attribute"
  ]
}

Hybrid Approach

Combine pattern matching and ML:

{
  "detection_method": "hybrid",
  "pattern_matching_weight": 0.4,
  "ml_classification_weight": 0.6
}

Intent-Specific Ranking

Apply different ranking strategies per intent:

{
  "intent_strategies": {
    "brand": {
      "boost_exact_matches": true,
      "diversity": "low",
      "guardrails": false
    },
    "category": {
      "boost_exact_matches": false,
      "diversity": "high",
      "guardrails": true,
      "max_per_brand": 2
    },
    "attribute": {
      "filter_by_attribute": true,
      "boost_attribute_match": true,
      "diversity": "medium"
    }
  }
}

Ambiguous Intent Handling

Handle queries with ambiguous or multiple intents:

{
  "ambiguous_intent_strategy": {
    "detection_threshold": 0.6,
    "fallback": "category",
    "show_diverse_results": true
  }
}

Intent Analytics

Track intent distribution and performance:

  • Intent distribution: What percentage of queries are each intent type?
  • Conversion by intent: Which intents convert best?
  • Intent accuracy: How accurate is intent detection?
  • Intent-specific metrics: Performance metrics per intent type

Best Practices

  1. Start with common intents: Focus on brand, category, and attribute intents first
  2. Validate detection: Regularly review intent detection accuracy
  3. Test strategies: A/B test intent-specific ranking strategies
  4. Handle ambiguity: Have fallback strategies for ambiguous queries
  5. Monitor performance: Track how intent handling affects conversion
  6. Iterate: Continuously improve intent detection and ranking strategies