What problem does it solve? Editorial teams collecting hotspots from RSS, Reddit, GitHub, and web sources need a consistent way to categorize, decompose, and pre-score every item before deciding what to write about. Doing this manually across dozens of daily items is slow and inconsistent. ## Core Features & Use Cases - Semantic Tagging: Assigns each hotspot a category, market scope, China relevance, risk level, and an overall score based on the caller-provided account context. - Event Decomposition: Breaks each item into normalized event parts (who, what, where, when, action type, object, occasion) so duplicate reports of the same event can be clustered downstream. - Multi-Dimensional Pre-Scoring: Outputs sub-scores for conflict, audience relevance, information gain, emotion, timeliness, impact, and source reliability, plus scoop, saturation, and duplicate penalties. - Use Case: A WeChat official account editor ingests 50 RSS items each morning, runs this stage to get tagged and scored JSON, then filters the topic pool by audience relevance and risk before the editorial meeting. ## Quick Start Provide a batch of collected hotspots with your account context and ask the AI to tag and pre-score every item into the structured JSON output.
