Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.bytestack.com/llms.txt

Use this file to discover all available pages before exploring further.

When you query multiple sources, the same real-world entity — a person, a company, a location — often appears in several places at once. A VP of Engineering might show up on LinkedIn with her job title, on GitHub with her repositories, and on Facebook with her city. Without any extra work on your part, ByteStack’s data stitching automatically detects that these records refer to the same individual and merges them into a single unified profile. This means you get a richer, more complete picture of each entity instead of fragmented, siloed records.

​
How entity resolution works

ByteStack’s stitching engine compares records across sources and looks for shared signals — names, usernames, profile photos, URLs, and contextual clues — to determine whether two records represent the same real-world entity. When confidence is high, ByteStack merges the records and attaches the original source data as linked profiles. Stitching runs automatically on query results that span multiple sources. You do not need to configure it or call a separate endpoint.

​
Supported entity types

People

Individuals identified across social profiles, professional networks, developer platforms, and news mentions. Stitching captures names, roles, affiliations, and activity.

Organizations

Companies, brands, and institutions that appear across LinkedIn, Google, news sources, and social media. Stitching unifies mentions, official pages, and third-party references.

Locations

Cities, venues, and geographic entities mentioned across platforms. Stitching connects place names, addresses, and location tags into a single canonical record.

​
Example: a person unified across four sources

Consider a query that surfaces Sarah Chen across LinkedIn, Google, Facebook, and GitHub. Without stitching, you would receive four separate records. With stitching, ByteStack identifies they all refer to the same person and returns a unified profile. Raw records before stitching:
SourceKey data
LinkedInVP Eng @ Acme, 47 posts, Technology industry
GoogleForbes Tech 40, speaker at DevConf 2026, 12 article mentions
FacebookLives in San Francisco, 142 mutual friends, 8 groups joined
GitHub28 public repos, 3.2K followers, 14K total stars
Unified profile after stitching: 
{
  "entity_id": "ent_sarah_chen_001",
  "entity_type": "person",
  "name": "Sarah Chen",
  "role": "VP Engineering",
  "employer": "Acme",
  "location": "San Francisco, CA",
  "sources": {
    "linkedin": {
      "headline": "VP Eng @ Acme",
      "industry": "Technology",
      "posts": 47
    },
    "google": {
      "title": "Forbes Tech 40",
      "mentions": 12,
      "speaking_events": ["DevConf 2026"]
    },
    "facebook": {
      "location": "Lives in SF",
      "mutual_friends": 142,
      "groups_joined": 8
    },
    "github": {
      "public_repos": 28,
      "followers": 3200,
      "total_stars": 14000
    }
  }
}
The top-level fields (name, role, employer, location) represent ByteStack’s resolved canonical values. The sources object preserves the original per-platform data so you can trace exactly where each field came from.

​
Unified profile fields

FieldDescription
entity_idUnique identifier for the resolved entity
entity_typeperson, organization, or location
nameCanonical resolved name
rolePrimary role or title (people only)
employerAffiliated organization (people only)
locationResolved geographic location
sourcesObject containing per-platform raw data
Data stitching significantly improves signal quality in research queries. Instead of counting the same person as three separate mentions across LinkedIn, Google, and GitHub, ByteStack counts them once and enriches the record with cross-platform context. This reduces noise and makes aggregated metrics more accurate.

​
When stitching applies

Stitching runs automatically when your query spans two or more sources and returns records that share identifiable entity signals. If ByteStack cannot resolve a match with sufficient confidence, it returns the records separately rather than risk an incorrect merge. You can always inspect the sources object on a unified profile to review the raw evidence used to stitch the entity.