Entity Resolution Engine

SecurePoint Trade

Four layers of intelligence. One confidence score.

Most screening tools do string matching. SecurePoint Trade combines phonetic, distance, similarity, and segment analysis to catch what simple searches miss — then tells your analysts exactly how confident the match is.

Phonetic matching catches sound-alike names across transliterations.

Levenshtein distance detects typos and character transpositions.

Jaro-Winkler similarity handles reordered name segments.

Trigram analysis scores partial overlaps in long entity names.

Entity Resolution Engine

Input entity

MOHAMMAD AL-RASHEED

Phonetic

Muhammed El-Rashid

89%

Levenshtein

Mohamad Al-Rasheed

94%

Jaro-Winkler

Mohammed Rasheed

91%

Trigram

Al Rasheed Mohammad

87%

Composite confidence

92%

Review Required

4

Matching Algorithms

Phonetic, distance, similarity, trigram

0–100%

Confidence Score

Single unified output

Sub-second

Resolution Time

Per counterparty entity

100%

Explainability

Every score broken down

Algorithm Breakdown

How each layer contributes to the final score

Each algorithm targets a different class of name variation. Together they produce a match confidence that simple screening cannot achieve.

Phonetic Matching

Sound-alike detection

Catches names that sound identical but are transliterated differently across languages and scripts.

"Muhammad" = "Mohammed" = "Muhammed"

Levenshtein Distance

Typo detection

Measures character insertions, deletions, and substitutions to catch data-entry errors and fat-finger typos.

"Al-Rasheed" → "Al-Rashead" (1 swap)

Jaro-Winkler

Transposition detection

Optimized for short strings and name prefixes. Catches character transpositions and partial matches.

"Rasheed" → "Rasehd" (transposed)

Trigram Similarity

Segment reordering

Breaks names into 3-character segments and compares overlap. Catches reordered name components.

"Al Rasheed Mohammad" ↔ "Mohammad Al-Rasheed"

All four algorithms converge into

One Confidence Score

A single 0-100% Match Confidence score tells your analysts exactly how likely a hit is to be real — reducing false positives and accelerating clear decisions.

Score Transparency

Every match score is fully explainable

When a counterparty scores 92%, your analyst sees exactly which algorithms contributed and why. No black boxes.

Score Detail

Screened Entity

MOHAMMAD AL-RASHEED

TXN-2024-08841
PhoneticMuhammed El-Rashid
Weight: 22%89%
LevenshteinMohamad Al-Rasheed
Weight: 28%94%
Jaro-WinklerMohammed Rasheed
Weight: 26%91%
TrigramAl Rasheed Mohammad
Weight: 24%87%

Composite Confidence

92%

Review Required
Threshold: 85%

Why this matters

String matching misses 30-40% of true matches on transliterated names

Studies of sanctions screening accuracy show that simple string matching consistently fails on names transliterated from Arabic, Cyrillic, and Chinese scripts. Entity resolution with multi-algorithm scoring closes this gap — catching the matches that matter while reducing the false positives that slow your team down.

4

Algorithms combined

92%

Sample composite score

< 1s

Per-entity resolution

FAQ

Entity resolution questions

For compliance teams evaluating screening accuracy and false positive reduction.

See entity resolution in action

Request a demo to see how 4-layer matching reduces false positives and catches what simple screening misses.