Make sense of imperfect data.
MatchForge brings string similarity, text preprocessing, record matching, blocking, and dataset deduplication together in one framework-independent toolkit.
Source
Jane Smith
jane@example.com
Target
JANE SMYTH
jane@example.com
Potential matching records
Similarity-based entity resolution
Conceptual illustration of the MatchForge workflow
How it works
From inconsistent records to meaningful connections.
Entity resolution involves more than comparing strings. MatchForge combines preprocessing, blocking, similarity algorithms, and record matching in one toolkit.
name
JORDAN LEE
JORDAN@EXAMPLE.COM
city
Ahmedabad
name
Jordan Leigh
jordan@example.com
city
Ahmedabad
The challenge
Differences in spelling, formatting, and missing information make direct equality checks unreliable.
Core capabilities
Everything you need to work with imperfect data.
Build entity resolution workflows using modular capabilities designed to work together.
Text preprocessing
Normalize casing, whitespace, punctuation, and accents before comparing strings.
String similarity
Compare strings using edit-distance, character-based, q-gram, and phonetic algorithms.
Record matching
Match records across datasets with configurable fields, weights, and thresholds.
Dataset deduplication
Identify duplicate records, group related entries, and retrieve unique records.
Candidate blocking
Generate candidate pairs using configurable blocking rules before matching.
Evaluation
Evaluate matching predictions and blocking results against labelled ground truth.
Work with real datasets
CSV workflows, built in.
Work directly with CSV data for matching, deduplication, and evaluation.
Built for developers
Start with one package.
Build your own workflow.
Use MatchForge Core in your JavaScript or TypeScript application. Its matching logic is independent of any frontend framework.
Import what you need
import {
normalizeString,
compareStrings,
matchRecords,
generateCandidates,
} from "@matchforge/core";