Events
An event is a normalized, source-linked observation that NataPulse has accepted into its financial intelligence model.
A provider item is not automatically an event. It first passes through validation, normalization, entity resolution, duplicate checks, source assessment, scoring, and publication controls.
Event anatomy
Section titled “Event anatomy”A product-facing event can include:
- a title and cleaned description;
- occurrence and ingestion timestamps;
- source family and original source reference;
- entity type and entity key, such as a ticker, coin, wallet, protocol, or topic;
- importance and confidence;
- sentiment where applicable;
- cluster membership;
- cross-source evidence;
- related reports or impact analysis;
- structured domain-specific fields.
The precise fields depend on the source. A filing event may expose form and issuer information. A market event may expose price behavior. An on-chain event may expose addresses, value, block data, and labels. A quantitative event may expose anomaly, volatility, and quality information.
Event versus observation
Section titled “Event versus observation”An observation is what a provider returned. An event is what NataPulse can safely and consistently reason about.
This distinction allows the product to:
- remove provider-specific formatting;
- avoid treating duplicates as independent evidence;
- resolve a company name, cashtag, token name, or address into a canonical entity;
- preserve source provenance;
- compare evidence across domains;
- apply the same publication and scoring rules.
One event is not always enough
Section titled “One event is not always enough”A single event can be material, especially when it comes from a primary source such as a regulatory filing. However, isolated low-trust or ambiguous observations should be treated cautiously. NataPulse uses confidence, source reliability, corroboration, and cluster formation to make that uncertainty visible.
Public and unpublished events
Section titled “Public and unpublished events”Not every normalized record is published to the user-facing product. Events may be withheld when they fail quality, reliability, relevance, or publication requirements. Product pages consume curated read models rather than raw internal tables.
Where events appear
Section titled “Where events appear”Events are visible in Live Pulse, Event Explorer, Social Radar, Filings Desk, Market Desk, On-chain Desk, Wallet Intelligence, reports, Deep Research evidence, alerts, and Analyst citations.