Top News RSS
Importance-Ranked Top 10
1. Overview
Top News RSS clusters the past 24 hours of important news at the story level and delivers a ranked Top 10 by structural importance. Out of hundreds of articles published across 20+ media outlets, AI selects the 10 most important stories and ranks them.
Story-level Top 10: A ranking of stories (same event merged into one), not individual articles.
Important news only: Only articles classified as Level 1–2 by the news analysis pipeline are used.
Updated every hour: To keep the ranking current, a new edition is generated on the hour using the most recent 24-hour window.
16 languages: Every ranking is delivered simultaneously in 16 languages.
AI Insight included: Every item includes structured analysis across five sections: Key Point, Market Sentiment, Similar Past Cases, Ripple Effect, and Opportunities & Risks.
News RSS delivers every individual article. Top News RSS delivers the 10 most important stories from the past 24 hours.
2. How the Ranking Is Generated
The ranking takes articles that the news analysis pipeline has determined to be important (Level 1–2) as input, then applies story clustering and structural importance sorting to produce the output.
Input: Level 1–2 articles only
Only articles classified as Level 1–2 by the news analysis pipeline are used. These are pre-verified important news.
Story clustering
Multiple reports about the same event are merged into a single story. Clustering is based on same actor + same action + same object. If five outlets cover the same event, the ranking delivers it as one consolidated story.
Representative item selection
For each story, the article that best reflects the latest material update is selected as the representative item. The title, description, link, and pubDate exposed in the RSS correspond to this representative item.
Importance-based sorting
Sorting uses structural importance (Impact × Actionability × Transmission) as the primary criterion, and coverage volume (how many outlets reported it) as the secondary criterion. When structural importance is equal, stories validated by more outlets rank higher. Repeated coverage from the same outlet is capped to prevent ranking distortion.
Top 10 output
The top 10 stories from the sorted list are selected and assigned rank 1–10. If fewer than 10 stories exist, only the available stories are returned.
This ranking shares the same input pipeline and story clustering with Daily Market Update RSS. The difference is the output format: Daily Market Update delivers a five-section narrative briefing, Top News RSS delivers a ranked Top 10 list.
3. How RSS Fields Map to UI
This section shows how each RSS field is rendered in the actual NS3 app.
① Update time ("Update an hour ago")
lastBuildDate (channel level)
② Ranking badge (#1, #2, ... #10)
rank
③ Headline text
title
Tap/click destination
link (AI Insight page)
In the app UI, the Top News ranking displays only rank and title. Description, images, and coin tags are not shown in the ranking list. When a user taps an individual item, the full analysis is available on the AI Insight detail page via link. This design is optimized for scannability and priority ordering.
The RSS feed also includes description, media:content, mentionedCoins, and insight fields, so platforms can render summaries, images, coin tags, and AI Insight alongside the ranking as needed.
Example feed: https://ns3.ai
4. Validate the Data Yourself
Verify ranking quality directly before integration.
Validate the ranking
Ask any AI model to search for recent crypto news first, then compare it against the Top News ranking.
Example prompt:
Search for the most important crypto news from the past 24 hours. Then read the Top 10 ranking at the following URL and evaluate whether the ranking accurately reflects the most important events. https://api.ns3.ai/feed/news-ranking?lang=en
Translation quality
Provide a language-specific Top News RSS to any AI model and ask whether it reads naturally to local financial news readers.
Replace the language code to test any of the 16 supported languages:
en · zh-CN · zh-TW · ko · ja · ru · tr · de · es · fr · vi · th · id · hi · it · pt
Example prompt:
Read the Top 10 ranking at the following URLs, one in English and one in Korean. Compare the two and evaluate whether a Korean-speaking user would get the same level of information and the same ability to act on the news as an English-speaking user. https://api.ns3.ai/feed/news-ranking?lang=en https://api.ns3.ai/feed/news-ranking?lang=ko
5. RSS URL & Languages
Base URL
16 Language URLs
English
en
https://api.ns3.ai/feed/news-ranking?lang=en
简体中文
zh-CN
https://api.ns3.ai/feed/news-ranking?lang=zh-CN
繁體中文
zh-TW
https://api.ns3.ai/feed/news-ranking?lang=zh-TW
한국어
ko
https://api.ns3.ai/feed/news-ranking?lang=ko
日本語
ja
https://api.ns3.ai/feed/news-ranking?lang=ja
Русский
ru
https://api.ns3.ai/feed/news-ranking?lang=ru
Türkçe
tr
https://api.ns3.ai/feed/news-ranking?lang=tr
Deutsch
de
https://api.ns3.ai/feed/news-ranking?lang=de
Español
es
https://api.ns3.ai/feed/news-ranking?lang=es
Français
fr
https://api.ns3.ai/feed/news-ranking?lang=fr
Tiếng Việt
vi
https://api.ns3.ai/feed/news-ranking?lang=vi
ไทย
th
https://api.ns3.ai/feed/news-ranking?lang=th
Bahasa Indonesia
id
https://api.ns3.ai/feed/news-ranking?lang=id
हिन्दी
hi
https://api.ns3.ai/feed/news-ranking?lang=hi
Italiano
it
https://api.ns3.ai/feed/news-ranking?lang=it
Português
pt
https://api.ns3.ai/feed/news-ranking?lang=pt
lang is a required parameter. This feed has no other filter parameters.
6. Item Field Specification
This feed returns the latest single set of the Top 10 ranking. Maximum 10 items.
6.1 Standard RSS Fields
title
Meaning: Headline of the representative item
Type:
string(CDATA-wrapped)Use: Headline text in the ranking list
description
Meaning: Core summary of the representative item (1–4 sentences)
Type:
string(CDATA-wrapped)Use: Article preview snippet
link
Meaning: NS3 AI Insight page URL for the representative item
Type:
URL stringUse: Click destination for the AI Insight detail page
guid
Meaning: Unique item identifier
Type:
URL string,isPermaLink="false"Note: Each item has a unique value. Standard RSS guid-based deduplication works.
pubDate
Meaning: Original publish time of the representative item
Type: RFC 822/1123 format
Use: Article publish time display
media:content
Meaning: Social preview image URL
Type:
URL string,medium="image"Note: May be missing. Some sources do not provide preview images.
6.2 Top News Extended Fields
rank
Meaning: Story importance ranking position
Values:
1–10(1 = most important)Type:
integerUse: Ranking badge (#1, #2, ... #10), sort order
mentionedCoins
Meaning: Token symbols related to the representative item
Type:
string(CSV format, e.g.,BTC,ETH,SOL)Note: May be missing. If no token is explicitly mentioned, the value may be empty or the tag absent.
Parsing: Split by
,→ trim each item → normalize to uppercase
6.3 Channel-Level Fields
lastBuildDate
Meaning: Time when the ranking was generated (channel level)
Type: RFC 822/1123 format
Use: "Update 3 minutes ago" display. Used to judge ranking freshness. Distinct from each item's
pubDate(article publish time).
6.4 AI Insight Field
insight
Meaning: Full AI structured analysis
Type:
string(CDATA-wrapped, markdown format)Structure: Sections delimited by
##headings. All items include all five sections (input is Level 1–2 only).
Parsing guide: Split on the ## string to extract individual sections.
6.5 Fields Not Included
newsType and level are absent from this feed. All items are representative items of Level 1–2 stories, so priority is expressed through rank rather than a separate level field.
7. Error Handling & Edge Cases
7.1 Missing media:content (no image)
media:content (no image)Cause: The original source does not provide a social preview image.
Product: Use a default image/logo or a no-image layout.
Engineering: Treat missing
media:content@urlas null.
7.2 Missing or empty mentionedCoins
mentionedCoinsCause: No token is explicitly mentioned in the representative item.
Product: Hide the coin tag area.
Engineering: Allow both missing tag and empty string.
7.3 Fewer than 10 items
Cause: Fewer than 10 Level 1–2 stories exist within the 24-hour window (weekends, holidays).
Product: Display only the available items. A natural list rendering is preferable to a "N of 10" label.
Engineering:
rankis sequential starting from 1, so item count can be determined directly.
7.4 pubDate vs lastBuildDate
pubDate vs lastBuildDatepubDate: Original publish time of the representative article (item level)lastBuildDate: Generation time of the ranking set (channel level)In UI, use
lastBuildDatefor "Update N minutes ago" andpubDatefor individual article timestamps.
7.5 HTTP error handling
200
Success
Process feed normally
304
Not Modified
Use cached version (when conditional request headers are supported)
400
Bad Request (e.g., invalid lang code)
Check parameter values
500, 502, 503
Server error
Retry after 30–60 seconds. Do not retry immediately in a tight loop.
If the server is unreachable or returns a non-200 response, continue serving the last successfully fetched ranking until the next successful poll.
8. Polling & Caching
Generation cycle: A new ranking is generated every hour on the hour.
Recommended polling window: Between minute 15 and minute 20 of each hour. Generation may be delayed when important news volume is high, so polling 15–20 minutes after the hour is more reliable than polling immediately.
Feed scope: Only the latest single ranking set is returned.
Translation delay: Translated rankings may arrive approximately 5–10 minutes after the English ranking is generated. When using non-English languages, adjust the polling window to minute 25–30.
9. Timezone & Display
lastBuildDateandpubDateare provided in GMT (RFC 822/1123).Convert to local time for UI display, but keep sorting and freshness checks anchored to UTC.
Calculate "Update N minutes ago" based on
lastBuildDate.
10. Implementation Checklist
11. FAQ
Q: How is this different from News RSS?
News RSS delivers every individual article in real time. Top News RSS clusters the past 24 hours of important articles (Level 1–2) at the story level and delivers the Top 10 by importance. Multiple reports about the same event are merged into one story, so readers can immediately see the 10 most important events without duplication.
Q: What is the relationship with Daily Market Update RSS?
They share the same input pipeline (Level 1–2 articles) and story clustering. The difference is the output format: Top News RSS delivers a ranked Top 10 list, Daily Market Update delivers a five-section narrative briefing.
Q: Why "Top 10 stories" instead of "Top 10 articles"?
If five outlets cover the same event, five articles exist. Ranking by article would fill the top positions with the same event. Clustering at the story level removes this duplication, ensuring the Top 10 shows 10 genuinely distinct events.
Q: Can two items have the same rank?
No. Rank is assigned sequentially starting from 1. Even in cases of tied scores, deterministic tiebreak rules ensure every story receives a unique rank.
Q: Can fewer than 10 items be returned?
Yes. If fewer than 10 Level 1–2 stories exist within the 24-hour window, only the available stories are returned. This is more likely on weekends and holidays when news volume is lower.
Q: What is the difference between pubDate and lastBuildDate?
pubDate is the original publish time of each representative article. lastBuildDate is the time the entire ranking set was generated. In UI, use lastBuildDate for "Update N minutes ago" and pubDate for individual article timestamps.
Q: Can translations be delayed?
Translated rankings may arrive approximately 5–10 minutes after the English ranking is generated. When using non-English languages, poll at minute 25–30 of each hour.
Last updated