AI Classification

Overview

Every article ingested by Rushed goes through a deep AI classification pipeline that extracts structured intelligence from unstructured news. This page explains every dimension of classification.

Sectors (25)

Every article is classified into a primary sector:

Code	Sector	Code	Sector	Code	Sector
`FIN`	Finance	`TRA`	Transportation	`FIP`	Private Equity
`TEC`	Technology	`COM`	Communications	`CIE`	Science
`SAU`	Healthcare	`SER`	Services	`DEF`	Defense
`ENE`	Energy	`GOV`	Government	`CUL`	Culture
`MAT`	Materials	`ESP`	Sports	`EDU`	Education
`IND`	Industrial	`CRY`	Crypto	`AMB`	Environment
`CON`	Consumer	`AGR`	Agriculture	`SOC`	Social
`IMO`	Real Estate			`TUR`	Tourism

Topics (28)

Articles are tagged with one or more topics describing the subject matter:

Code	Topic	Description
`M&A`	Mergers & Acquisitions	Buyouts, sales, corporate incorporations
`IPO`	Public Offerings	IPOs, follow-ons, delistings
`RES`	Financial Results	Quarterly/annual earnings, revenue, balance sheets
`DIV`	Dividends	Dividend distributions, interest on equity
`REG`	Regulation	Laws, rules, oversight, compliance
`POL`	Politics	Elections, government, congress, judiciary
`MAC`	Macroeconomics	GDP, inflation, interest rates, FX, employment
`ESG`	ESG / Sustainability	Environmental, social, and corporate governance
`INO`	Innovation / R&D	New products, patents, research & development
`CRI`	Crisis / Scandals	Fraud, recalls, accidents, reputation
`EXE`	Executives / Leadership	CEO changes, board shifts, senior management
`TRB`	Labor	Strikes, mass layoffs, labor agreements
`INT`	International	Foreign trade, geopolitics, sanctions
`JUR`	Legal	Lawsuits, court rulings, arbitration
`EXP`	Expansion / Contraction	Opening/closing of units and markets
`PAR`	Partnerships / Contracts	Commercial agreements, joint ventures
`DEB`	Debt / Credit	Bond issuances, renegotiations, credit ratings
`TRF`	Sports Transfers	Player signings, loans, contract clauses
`CPT`	Championships	Leagues, tournaments, standings, playoffs
`INF`	Infrastructure	Construction, railways, roads, utilities, PPPs
`CLI`	Climate / Extreme Events	Floods, droughts, hurricanes, wildfires
`SEG`	Security / Crime	Police operations, trafficking, federal investigations
`SAP`	Public Health	Pandemics, vaccination, epidemics, public health policy
`ELE`	Elections	Candidates, polls, electoral courts, campaigns
`TRI`	Tax / Fiscal	Tax reform, duties, ICMS, VAT, exemptions
`CIB`	Cybersecurity / Privacy	Hacks, data breaches, LGPD, ransomware
`EDI`	Editorial / Opinion	Editorials, columns, expert analysis

Content types (7)

Each article is classified by content type with an impact multiplier:

Type	Multiplier	Description
`BREAKING`	2.0x	Breaking news, first report
`REPORT`	1.5x	In-depth report or investigation
`ANALYSIS`	1.2x	Expert analysis or opinion piece
`INTERVIEW`	1.1x	Interview with source
`NEWS`	1.0x	Standard news article
`PRESS`	0.8x	Press release, corporate communication
`RUMOR`	0.5x	Unconfirmed, speculative

Urgency levels (4)

Level	TTL	Description
`CRITICAL`	15 min	Market-moving, requires immediate attention
`HIGH`	2 hours	Significant, time-sensitive
`MEDIUM`	12 hours	Relevant, not urgent
`LOW`	48 hours	Background, reference

Sentiment analysis (5 levels)

Label	Range	Confidence
`MUITO_POSITIVO`	0.7 to 1.0	High confidence required
`POSITIVO`	0.3 to 0.69
`NEUTRO`	-0.29 to 0.29
`NEGATIVO`	-0.69 to -0.3
`MUITO_NEGATIVO`	-1.0 to -0.7	High confidence required

Market impact (4 dimensions)

Every article is scored on its potential market impact:

Dimension	Values
Level	`ALTO`, `MEDIO`, `BAIXO`, `NENHUM`
Horizon	`IMEDIATO` (<24h), `CURTO` (1–7d), `MEDIO` (1–3m), `LONGO` (>3m)
Scope	`EMPRESA`, `SETOR`, `MERCADO`, `ECONOMIA`
Confidence	`CONFIRMADO`, `PROVAVEL`, `ESPECULATIVO`

5-Dimension scoring (0–100)

Each article receives a composite score across 5 dimensions:

Dimension	Range	Description
Source Score	5–25	Source credibility tier (TIER_1 = Bloomberg, Reuters = 21–25)
Depth Score	3–15	Investigation depth vs. flash news
Exclusivity Score	0–20	Scoop vs. republication
Financial Impact Score	0–40	Potential market impact
Type Multiplier	0.6–2.0x	Content type multiplier

Entity extraction (10 types)

Rushed extracts structured entities from every article:

Type	Fields
`COMPANY`	name, ticker, CNPJ
`PERSON`	name, role
`INSTITUTION`	name
`TICKER`	symbol
`LOCATION`	name
`EVENT`	name
`PRODUCT`	name
`LAW`	name, number
`MONETARY`	value, currency
`METRIC`	value, unit

For each entity: mention count, prominence score, sentiment toward entity, is_in_title, is_in_lead.

Bilingual translation

Every article is translated in both directions:

Language	Tone	Fields
PT	Editorial, direct — clear and objective	title, subtitle, summary
EN	Punchy, conversational — accessible and engaging	title, subtitle, summary

Getting Started

Core Concepts

AI Classification

Overview

Sectors (25)

Topics (28)

Content types (7)

Urgency levels (4)

Sentiment analysis (5 levels)

Market impact (4 dimensions)

5-Dimension scoring (0–100)

Entity extraction (10 types)

Bilingual translation

​Overview

​Sectors (25)

​Topics (28)

​Content types (7)

​Urgency levels (4)

​Sentiment analysis (5 levels)

​Market impact (4 dimensions)

​5-Dimension scoring (0–100)

​Entity extraction (10 types)

​Bilingual translation

Overview

Sectors (25)

Topics (28)

Content types (7)

Urgency levels (4)

Sentiment analysis (5 levels)

Market impact (4 dimensions)

5-Dimension scoring (0–100)

Entity extraction (10 types)

Bilingual translation