Analysis updated 2026-05-18
Test whether a privacy or PII-extraction system finds the right facts in generated documents.
Benchmark an entity-resolution system against records with realistic name confusion.
Score a relationship-inference system against known ground truth.
Generate safe, fictional test data instead of using real personal information.
| bluntmachetti/synthworld | 0xallam/my-recipe | 0xhassaan/nn-from-scratch | |
|---|---|---|---|
| Stars | 0 | — | 0 |
| Language | Python | Python | Python |
| Last pushed | — | 2022-11-22 | — |
| Maintenance | — | Dormant | — |
| Setup difficulty | easy | moderate | moderate |
| Complexity | 3/5 | 2/5 | 4/5 |
| Audience | developer | general | developer |
Figures from each repo's GitHub metadata at analysis time.
Installable via pip, browsing pre-generated benchmarks online needs no install at all.
SynthWorld is a tool for generating fake but realistic populations of people, complete with family, friend, and coworker connections, for the purpose of testing privacy and identity software. The project explains its role with a simple comparison, a general fake-data generator creates individual rows of information, while SynthWorld creates connected worlds of people along with a hidden answer key describing exactly which facts are true. The generated identities are deliberately made confusing in realistic ways, including people who share common names, use different spellings of the same name, or have nicknames, so that systems being tested have to work through genuine ambiguity rather than clean, easy data. Because everything is generated from a starting seed number, the same command always produces the exact same population, which makes results repeatable and comparable over time. The tool is aimed at teams building or evaluating software that extracts personal information from documents, matches records to the same person, infers relationships between people, or scores how exposed someone's personal data is. It ships as a command line tool that can generate a population of people, generate related test scenarios, and grade a system's output against the hidden answer key using a separate evaluate command. A public, already-generated set of benchmark data is also available to browse online without installing anything. SynthWorld grew out of a companion project called Idcognito but is designed to work as its own independent piece of software. It explicitly does not turn real, sensitive data into anonymized data, it only creates fictional data from scratch. It is written in Python, distributed as a package called idcognito-synthworld, and released under the Apache-2.0 license, which allows free use, modification, and redistribution, including for commercial purposes.
A tool that generates connected, fictional populations of people with an answer key, for testing privacy, entity-matching, and identity software.
Mainly Python. The stack also includes Python, CLI.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated easy, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.