News Intelligence — Real-Time RSS News Analysis & Intelligence Outputs

News Intelligence is a production-ready Apify Actor that ingests RSS and Atom news feeds, normalizes articles, and transforms high-volume news streams into structured, intelligence-ready datasets.

The Actor is designed to separate raw signal from actionable intelligence, enabling downstream systems (alerts, analytics, AI workflows) to operate on reliable data rather than noisy feeds.

---

What This Actor Does

This Actor converts RSS news feeds into automation-ready intelligence by:

• Ingesting RSS / Atom feeds and approved news listing URLs
• Normalizing article metadata across sources
• Filtering articles using a configurable time window
• Computing confidence scores and summary metrics
• Emitting clean, deterministic datasets for downstream use

⚠️ The Actor does not send alerts directly.
It emits alert-ready intelligence signals that are intended to be consumed by external systems.

---

Typical Workflow

1. Create a task with one or more RSS feeds
2. Schedule the task (hourly, daily, etc.)
3. The Actor ingests and normalizes articles
4. Structured datasets are generated
5. Consume outputs via API, dashboards, or automation tools
6. Implement alerts externally (Zapier, webhooks, BI tools, ML pipelines)

This keeps the Actor lightweight, scalable, and deterministic.

---

Input Configuration

Sources

RSS feed URLs or approved news listing URLs.
One URL per line.

Example: https://techcrunch.com/feed/

https://www.reuters.com/rssFeed/

Pipeline Stage

Controls internal processing stages:

• ingest — article ingestion only
• cluster — topic grouping
• intelligence — full pipeline (default)

Changing the stage does not change output dataset structure.

Keyword Controls

• Include keywords — prioritize matching content
• Exclude keywords — ignore matching content
• Priority keywords — boost confidence scoring

Time Window (Hours)

Only include articles published within this time window (if timestamps exist).

Max Articles

Maximum number of articles processed per run.

Alert Signal Thresholds

Used only for generating alert-ready intelligence signals:

• Alert spike threshold
• Minimum confidence for alert

⚠️ Alerts must be implemented externally.

---

Output Datasets

The Actor emits three stable datasets, exposed via the output schema.

---

Articles Dataset

Each item represents a normalized news article.

Example Article Item

{
  "title": "OpenAI releases new reasoning model",
  "url": "https://www.reuters.com/technology/openai-releases-new-model",
  "source": "Reuters",
  "publishedAt": "2025-12-29T09:14:00Z",
  "confidence": 0.82
}

Use Cases

This Actor is designed to support the following intelligence and analytics workflows:

• Coverage spike detection
• Trend analysis
• Alert threshold evaluation
• Machine learning feature inputs

The structured outputs allow downstream systems to operate on reliable signals rather than raw news volume.

---

Alert System (Design & Roadmap)

⚠️ Alert logic is not executed by the Actor in the current version.

The Actor emits alert-ready intelligence signals only.
All alerting and notification logic must be implemented externally.

---

Planned Alert Signals

The following alert types are part of the planned intelligence model and can be derived externally:

• NEW_TOPIC_EMERGED
• SPIKE_IN_COVERAGE
• SENTIMENT_SHIFT

---

Confidence-Aware Gating (Planned)

Alert signals are intended to be acted on only when the following condition is met:

This approach significantly reduces false positives and alert fatigue in downstream systems.

---

Pricing

This Actor uses pay-per-event pricing.

You are charged a fixed fee each time the Actor starts running.
The total cost depends on the Actor memory configuration (one event per GB, minimum one event).

This pricing model makes costs predictable for scheduled intelligence and monitoring workflows.

---

Version History

v0.2 — Intelligence-Ready Outputs (Current)

• Multi-dataset architecture
• Confidence scoring
• Time-window filtering
• Overview and metrics datasets
• Alert-ready intelligence model (externalized logic)

v0.1 — Article Collection

• RSS ingestion
• Normalized article metadata
• Deduplication

---

Summary

• Converts RSS news feeds into structured intelligence
• Designed for analytics, alerts, and AI workflows
• Deterministic, scalable, and automation-friendly
• Alert logic intentionally externalized
• Marketplace-compliant and production-ready