Product Group Taxonomy

The Product Group Taxonomy helps improve search results for generic search terms by linking them to specific product names.

Problem: Poor Search Results for Generic Terms

Certain products do not contain generic names in their descriptions. For example, “bicycles” are often listed under names like City-Bike, Road Bike, or MTB, while the only items containing the word “bicycle” are accessories. This results in unsatisfactory search results for the term “bicycle.”

A similar issue occurs with “shoes” and other generic terms.

Why Not Use Synonyms?

1. Too Many Synonyms Needed

• Instead of searching for just “bicycle”, a synonym list would require terms like City-Bike, MTB, ATB, etc.
• Searching for 10–20 terms at once negatively impacts search performance.

2. Synonym Ranking Issue

• Synonyms have a built-in ranking penalty.
• This means that original words (e.g., “bicycle”) will always rank higher, pushing actual bicycles behind accessories.

Note

While synonyms are useful, they become less effective when handling broad generic queries with many variations and subtypes.

Solution: Product Group Taxonomy

The Product Group Taxonomy overcomes these limitations by organizing synonyms in a tree structure. For example:

• Instead of mapping MTB → Bicycle, the hierarchy is: Bicycle → Sports Bike → MTB

How Performance is Improved

• Synonyms are added directly to product data, reducing the need for complex multi-term searches.
• When a leaf term (the most specific category) appears in a product, all parent terms in the hierarchy are automatically added.

Example: A product tagged as MTB is automatically linked to Sports Bike and Bicycle. This ensures relevant products appear higher in search results for broad terms.

Tip

This approach ensures that generic searches return actual products—not just accessories or loosely related items.

How the Taxonomy is Applied

Since taxonomy updates require changes in product styles and the search index, they do not take effect immediately. All changes are processed once per night.

Processing Steps

1. The system checks if the commodity group field (commodityGroup) contains a leaf term. If found, all related parent terms in the hierarchy are added automatically.
2. If the commodity group field is empty, the system applies the process to the Short Description field: It extracts the first part of the description (before the first comma, if present). Each word is then analyzed, and matching taxonomy terms are appended to the Short Description field.