超越分类：AI代理如何在规模上演变Shopify的产品分类法

Blog post written by Kshetrajna Raghavan and Ricardo Tejedor

博客文章由Kshetrajna Raghavan和Ricardo Tejedor撰写

Our product classification system processes over tens of millions of predictions daily with a high degree of accuracy. But what happens when the taxonomy that powers it needs to grow and adapt? We needed a solution that ensures that our 10,000+ categories and 2,000+ attributes continue to serve merchants and customers as commerce evolves.

我们的产品分类系统每天处理超过数千万的预测，准确度极高。但是，当支撑它的分类法需要增长和适应时会发生什么？我们需要一个解决方案，以确保我们的10,000多个类别和2,000多个属性在商业演变中继续服务于商家和客户。

The solution: an innovative AI multi-agent system that goes beyond classifying products—one that actively improves the Taxonomy labels themselves, keeping our system agile and future-proof.

解决方案：一个创新的AI多代理系统，超越了产品分类——一个积极改善分类标签本身的系统，使我们的系统保持灵活和未来-proof。

The challenge: Keeping taxonomy current at scale

挑战：在规模上保持分类的最新性

The taxonomy that powers our product classification system faces its own scaling challenges. Commerce never stands still: new products emerge, existing categories evolve, and merchant needs shift as markets change.

推动我们产品分类系统的分类法面临自身的扩展挑战。商业从不静止：新产品不断出现，现有类别不断演变，随着市场变化，商家的需求也在变化。

The volume problem

数量问题

Managing a product taxonomy on a worldwide scale requires constant attention. Every new product type, emerging technology category, and seasonal trend potentially requires taxonomy updates. Traditional manual curation simply can't keep pace.

在全球范围内管理产品分类法需要持续关注。每种新产品类型、新兴技术类别和季节性趋势都可能需要更新分类法。传统的手动策划根本无法跟上步伐。

For example, consider the rapid emergence of categories like smart home devices, sustainable products, or remote work equipment. Each category represents not just new categories but also entirely new attribute sets.

例如，考虑智能家居设备、可持续产品或远程工作设备等类别的快速出现。每个类别不仅代表新的类别，还代表全新的属性集。

Smart home devices need connectivity types, power requirements, and compatibility– specifications that didn't exist in our taxonomy before.

智能家居设备需要连接类型、电力要求和兼容性——这些规格在我们的分类法中之前并不存在。

The expertise problem

专业知识问题

Effective taxonomy design requires deep domain expertise. Understanding the nuance...