> ## Documentation Index
> Fetch the complete documentation index at: https://docs.qoory.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Data collection and normalization

> How Qoory turns varied market, entity, news, and social inputs into research data.

Qoory's data pipeline separates collection, normalization, identity resolution,
enrichment, search indexing, and presentation. This lets each stage be checked
without treating one source response as a finished research record.

## 1. Collect

Scheduled pipelines collect structured entity data, market and DeFi
observations, public filings, news, social evidence, GitHub activity, events,
funding, and selected on-chain data.

## 2. Normalize

Source-specific fields are converted into Qoory types, units, timestamps, and
nullable values. Only normalized fields needed by a product or documented API
response are presented to users.

## 3. Resolve

Records are matched to stable projects, tokens, people, funds, X accounts,
stocks, news clusters, narratives, or events. Ambiguous records can remain
unlinked until more evidence is available.

## 4. Enrich and connect

Qoory adds derived relationships and bounded classifications, such as a token's
project, a person's role, a fund's participation in a round, or a news story's
entity links.

## 5. Index and score

Search documents and ranking signals are refreshed from normalized records.
Search relevance, long-term ranking, and short-window momentum are different
mechanisms and should not be interpreted as one universal quality score.

## 6. Present with context

The app and API return purpose-built response shapes rather than complete source
or storage records. Each surface includes only the fields needed for the
documented research workflow.