Definition
Product unit normalization is the process of mapping source measurements into consistent quantity and unit fields while preserving their original meaning. A commerce ontology should distinguish a single item, a multipack, the quantity per item, and the total quantity rather than storing them as an ambiguous product-name fragment.
Why It Matters
- A catalog entry such as 2 x 250 ml describes two containers and 500 ml total. Treating the number 250 as a unitless attribute can produce incorrect comparisons.
- Agents and filters need compatible dimensions. A mass measurement cannot be treated as a volume measurement without valid product-specific information.
- Reliable quantity structure supports comparable offers, clearer product answers, and analysis of quantity-related dissatisfaction within a Commerce Intelligence OS.
How It Works
- Extract the source quantity, unit, pack count, measurement dimension, and provenance into separate fields. Preserve the original label for audit.
- Convert compatible units into a chosen canonical representation using explicit conversion rules. Flag ambiguous labels instead of guessing missing values.
- Calculate total quantity only when pack structure is known. Keep packaging size, usable product quantity, and dimensions separate.
- Validate normalized values against source data and sample customer-facing comparisons. Publish corrections to catalog consumers and inspect affected agent answers.
Ecommerce Example
Context: Illustrative example: one detergent listing contains two 250 ml bottles for 300 currency units, while another contains one 400 ml bottle for 260.
Recommended move: The normalized totals are 500 ml and 400 ml. Their prices per 100 ml are 60 and 65 respectively, assuming the same currency and comparable price basis.
Why it matters: The first offer has a lower price per volume in this example. That does not prove equal concentration, performance, or suitability; those remain separate product attributes.
iKawn Framework
Model
Represent count, quantity, unit, and product relationships explicitly in the iKawn ontology framework.
Normalize
Apply traceable conversions only within compatible dimensions.
Validate
Check source meaning and quarantine ambiguous pack descriptions.
Explain
Give agents both comparable quantities and the limitations of the comparison.
Concise Summary
Unit normalization makes quantities comparable without erasing pack meaning. Keep dimensions, provenance, and uncertainty explicit so catalog intelligence does not turn missing data into confident answers.