Xianyu Product Attribute Completion Strategy

Background : Detailed product attributes improve item presentation, guide buyers, and enhance recommendation efficiency. Structured key attributes enable better product layering and selection.

Current Issues : Xianyu’s lightweight publishing leads to insufficient attributes; second‑hand items are often unique, preventing reuse of Taobao categories; most sellers are individuals with limited professionalism.

Overall Strategy : Introduce a framework consisting of a MetaQListener (publish message listener), ConstructPropertyManager (attribute construction), and MetaQProducer (consumer). The system leverages Alibaba MetaQ for messaging.

Key Attribute Construction :

Algorithm‑based prediction: image recognition predicts category, similar items, and brand. High‑precision predictions directly fill attributes; otherwise, users are guided to select.

Offline configuration: for attributes not predicted, predefined options are stored via the Switch framework and presented to users.

Attribute Completion Path : Two entry points—image tagging and “more info” tags—are supplemented by two guiding mechanisms: a task system that rewards sellers for completing information, and predictive input suggestions during publishing.

Closed‑Loop Optimization : By analyzing sales metrics (CTR, turnover rate) of completed attributes, the system iteratively validates and refines key attributes, shortening verification cycles by 70% and increasing UV‑CTR by ~3%.

Application Scenarios : Enhanced attribute display in search, recommendation feeds, and product detail pages; richer search filtering and guided shopping.

Launch Results : Coverage of six major categories, 30% increase in structured attribute coverage, and faster attribute verification.

Summary & Outlook : Future work includes building a standard operation backend for attribute selection, expanding attribute usage in layering and search, and improving pricing for second‑hand items.