AI New Media Content Production Workflow Orchestrator

Problem It Solves

Content teams working on new media often do not lack hot topics; they lack a way to separate industry-relevant signals from generic platform noise. Baidu, Weibo, Douyin, and Zhihu rankings mix tech, entertainment, local events, and trending memes, so editors still need to manually decide which items matter for their business. AI New Media Content Production Workflow Orchestrator addresses this by turning multi-platform hot topic collection into a scored industry hotspot list, reducing manual screening work.

How It Works

The skill is built on hotspot-aggregator and adds multi-platform aggregation, industry keyword filtering, and relevance scoring:

• Load industry configuration: It first reads the industry keyword monitoring fields from the configuration manager, which are required for downstream filtering.
• Call the underlying collector: It covers Baidu hot search, Weibo hot search, Douyin trending, and Zhihu trending, limited to hotspots from the past 24 hours.
• Match keywords: It applies three matching rules with different weights: exact match gets 3, semantic association gets 2, and fuzzy match gets 1, helping distinguish strong relevance from weak relevance.
• Score and rank: It generates relevance scores and outputs a Top 10 industry hotspot list with heat scores and relevance scores.
• Fallback behavior: If the underlying collector fails, it can fall back to fetching Baidu hot search via web_fetch; if no strongly relevant hotspots are found, it outputs generic industry hotspots and marks them as lower relevance.

Scope and Caveats

This skill is best used for initial industry hotspot filtering, not for final topic selection or content drafting. A few constraints matter in practice:

• Industry configuration must be loaded before execution.
• The keyword list should stay within 3 to 20 keywords to avoid diluting filtering precision.
• The output targets a ranked list: at least 3 items in the Top 10 should reach a high relevance threshold, defined as a relevance score of at least 7.
• It provides structured hotspot input, while fact-checking, editorial judgment, and content angle decisions still need to be handled by editors or a downstream agent workflow.