What is Entity Resolution?

Last updated: 26 October 2026

What is entity resolution?

Entity resolution is the process of identifying and linking multiple records that refer to the same real-world entity, whether a person, organization or company, across different data sources. In financial crime compliance, it means disambiguating and consolidating fragmented data into a single, accurate view of each party and the relationships between them.

Without it, the same customer can appear under several spellings, accounts or systems, and a risk that is obvious in one record stays hidden in another.

Why is entity resolution important in financial crime compliance?

Criminals rely on fragmented data. Name variations, shell companies, multiple accounts and layered ownership all make it harder to see that several records belong to one party. Entity resolution closes that gap and supports the fight against money laundering, terrorist financing and other illicit activity. Its main contributions are:

• Accurate data analysis: a single view of each party replaces duplicated and conflicting records.
• Identifying high-risk customers: hidden links to sanctioned parties, PEPs or known bad actors surface.
• Preventing fraud: shared identifiers such as phone numbers, addresses and devices expose coordinated activity, including payment fraud and money mule networks.
• Meeting regulatory requirements: supports KYC, AML and regulatory reporting obligations.
• Improving due diligence and monitoring: feeds CDD and ongoing monitoring with complete, current data.

How does entity resolution work?

1. Data collection. Records are gathered from internal systems, such as core banking, onboarding and payments, and from external sources, such as sanctions lists, registries and adverse media.
2. Data cleaning. Data is standardized, de-duplicated and corrected, including formats, abbreviations and character sets.
3. Feature extraction. Attributes such as name, date of birth, address, identifiers and contact details are extracted for comparison.
4. Entity matching. Records are compared using fuzzy matching and scored for the likelihood that they refer to the same entity.
5. Entity consolidation. Matched records are merged or linked into a single entity profile, keeping a history of where each attribute came from.
6. Entity verification. Results are validated through review and feedback, so the model improves over time.

What are the primary use cases of entity resolution?

• Customer due diligence: building a complete customer view across products and systems, including related parties and ultimate beneficial owners.
• Fraud detection: linking accounts and identities that share attributes, which is central to detecting synthetic identities and mule networks.
• Sanctions screening and name screening: resolving a name to the right entity, so alerts are accurate and decisions are applied consistently, including under the 50 percent rule.
• AML and transaction monitoring: assessing activity at the level of the real customer or network, not individual accounts.

What methods and models are used in entity resolution?

• Rule-based matching: deterministic rules, such as exact or near-exact matches on defined fields. Transparent and easy to explain, but rigid.
• Probabilistic matching: statistical scoring of the likelihood of a match across weighted attributes.
• Machine learning: models learn from labeled matches and analyst decisions, handling name variation and noisy data better than fixed rules.
• Hybrid models: combine rules, probabilities and machine learning, with rules providing explainability and models providing accuracy.

Graph-based approaches add a further layer, by resolving not just who an entity is but how it connects to others.

What are the benefits of entity resolution?

• Improved data quality: fewer duplicates and a single trusted view of each party.
• Enhanced compliance: consistent decisions and a clear audit trail.
• Operational efficiency: less time spent on manual matching and reconciling records.
• Enhanced KYC and CDD: more complete profiles and better risk assessment.
• Improved risk management: hidden relationships and criminal networks are exposed.
• Cost and productivity savings: analysts focus on genuine risk, with fewer false positives to clear.

What are the challenges in entity resolution?

• Data quality: incomplete, inconsistent or outdated records weaken every match.
• Scalability: matching across millions of records in near real time is computationally demanding.
• Data volume and complexity: multiple sources, formats, languages and scripts.
• Entity ambiguity: common names, shared addresses and similar businesses create false links.
• Privacy concerns: linking data across sources must respect data protection rules.
• The precision and recall trade-off: tightening or loosening a match threshold changes the balance of missed matches against false links.

In compliance, the cost of each error type differs. A missed match can hide a sanctioned party, while a false match can wrongly link or block an innocent customer. Thresholds should reflect the institution’s risk-based approach.

What are best practices in entity resolution?

• Start with high-quality data collection, with standardized formats and complete key fields.
• Update models regularly as data, name patterns and typologies change.
• Use advanced technology, including machine learning and AI, alongside transparent rules.
• Train teams to understand and challenge match decisions.
• Audit and validate results, keeping evidence of why entities were linked or separated.
• Keep explainability central, so every merge can be justified to a regulator.
• Apply one resolved view across controls, so screening, monitoring, KYC and case management all use the same entity.

For ownership structures in particular, see the FATF guidance on beneficial ownership.

How is AI changing entity resolution?

Machine learning improves matching on messy, multilingual data, and graph analytics reveal networks that record by record comparison cannot. Agentic AI goes further, by gathering evidence, resolving ambiguous matches and documenting the rationale for an analyst to confirm. Human oversight remains essential, so that every automated link can be explained and audited.

Introducing entity resolution from SymphonyAI

SymphonyAI’s entity resolution capability supports AML, payment fraud, KYC/CDD and sanctions screening, and is built into our financial crime prevention software. It resolves customer, counterparty and third party data into a single view, visualizes relationships and identifies shared contact details, so teams can see connected risk and focus on genuine threats. It underpins Symphony Risk Intelligence and Always-on Compliance.

Frequently asked questions

What is entity resolution in AML?
Entity resolution links records that refer to the same real-world person or organization across data sources, giving compliance teams one accurate view of each party and their connections.

What is the difference between entity resolution and data matching?
Data matching compares individual fields or records, while entity resolution is the full process of matching, consolidating and verifying records into a single entity profile.

Why is entity resolution important for sanctions screening?
It makes sure a name is resolved to the right entity, which reduces false positives and helps prevent missed matches, including those hidden behind ownership structures.

What techniques are used for entity resolution?
Rule-based, probabilistic, machine learning and hybrid approaches, often combined with graph analysis to understand relationships.

What is the precision and recall trade-off?
Precision measures how many proposed matches are correct, and recall measures how many true matches are found. Improving one usually reduces the other, so thresholds must be set deliberately.

Does entity resolution reduce false positives?
Yes. Resolving duplicates and linking related records means each party is screened and assessed once, with full context, rather than generating repeated alerts.