Tag Factory is a tag development and service platform for enterprise data teams. Use it to build tags from raw data, segment audiences for campaigns or risk control, and deploy tags as queryable services — all without managing separate pipeline infrastructure.
Prerequisites
Before you begin, ensure that you have:
- Purchased the Tag Factory value-added service and activated the Tag Factory module for your tenant.
Limits
Supported offline computing engines include MaxCompute, CDH6.x Hadoop, CDH5.x Hadoop, FusionInsight 8.x Hadoop, TDH 6.x, ArgoDB, and Cloudera Data Platform 7.x.
When using CDH6.x Hadoop, CDH5.x Hadoop, or TDH 6.x as the offline computing engine, only the open-source Flink real-time computing engine is supported for real-time tag development.
Feature overview
Tag Factory covers the full lifecycle of tag work: model entities and ID types, build and compute tags using offline or real-time engines, segment audience groups, and expose tags through data services.
Business planning
Entity creation: Tag Factory uses the Object-Link-Tag (OLT) model for entity modeling. Create business entities such as users or merchants, then bind them with ID types — user numbers, employee numbers, mailboxes — to map each entity across its different identifiers.
ID creation: Define the ID types associated with each entity, such as phone numbers and mailboxes for consumers and merchants.
Tag development
Dataset definition: Before building tags, define datasets from your data sources. A dataset preprocesses the source data — setting the timestamp range and selecting which columns to expose — so tag logic has a clean, scoped input.
Behavior relationship: Define interactions between entities through specific behaviors. For example, a user purchasing goods links the user and goods entities through the purchase behavior. These relationships feed behavior-statistics and preference-type tags, and can also drive audience group selection.
-
Tag processing: Tags describe characteristics of business entities. Tag Factory supports four tag types, each suited to a different computation model and use case:
Tag type
How it computes
When to use
Offline
Batch jobs on the offline computing engine (for example, MaxCompute)
Stable attributes updated on a schedule, such as a user's purchase count in the past 30 days
Real-time
Stream processing via the real-time computing engine
Attributes that must reflect the latest event, such as a user's most recent login timestamp
Service
Served directly from a data service layer without a batch or stream job
Low-latency lookups against pre-computed results
Advanced
Custom logic combining multiple computation models
Complex derived attributes that do not fit a single computation pattern
Audience group selection
Combine tag filters and behavior relationships to segment entities into audience groups. For example, classify users under 18 as the minor user group for targeted campaigns or compliance workflows. Use these groups for downstream analysis or as inputs to service applications.
Service application
Asset Market: Publish tags and audience groups developed on the workbench so other teams can discover and request them.
Asset Application: Manage tags and groups your team has requested from the Asset Market, along with any services built from them.
Access Tag Factory
On the Dataphin home page, click Tag in the top menu bar. The default landing page is the Asset Market, where you can browse and request published tags and audience groups.