Analysis of Notarial Operations Networks


Improve the detection of potential money laundering operations through the analysis of networks of people linked through notarial operations.

An organization specialized in the detection of possible money laundering and terrorist financing operations from notarial operations seeks to improve and complement its methodology.

There are more than 2,300,000 legal entities registered in its database, and in the period 2005-2017, they issued more than 5,000 notifications revealing hints of money laundering or terrorism financing. They already have business rules in place to detect it but are further looking for a new approach to help them improve their performance.

The IIC proposes a project based on the analysis of social networks. From the data of the notarial operations carried out, relationships between documents and legal entities can be revealed to guide the detection of money laundering. Specifically, two solutions are proposed: detection of clusters and anomalies.


The first step in a social network analysis project is to define relationships, and what defines a relationship between two legal entities is having participated in one or more notarial operations. With this idea, after pre-processing two years of operations data, 100 million relationships and 47 million nodes of all types were obtained, such as documents, legal entities, objects, and more.

Based on this information, we worked on the two proposed solutions:

Detection of clusters of legal entities and notarial documents

Communities or groups of related notarial operations are detected, as well as the legal entities involved.

Besides, for each cluster, a graph of relations and their characteristics is provided, so that the organization knows where to continue investigating if it detects a case of money laundering.

Detection of anomalies in notarial documents

A machine learning model was developed to detect anomalous patterns.

As input to the algorithm, almost 14 million notarial documents were analyzed, and more than 400 variables, traditional and based on the network approach, were taken into account.  How many documents is this notarial operation related to? What is the position of this document in the network? With all this, the algorithm provides an anomaly score, between 0 and 1, to each notarial document, ordering them from least to most anomalous.


Social network analysis technology allows us to relate the information that is already available and provide a new approach and solution. The network approach allows to structure and analyze information using a different logic to tabular databases (SQL) and, consequently, detect relationships that would otherwise go unnoticed. A relational approach is very useful in the context of money laundering.

The anomaly score, in turn, provides an analysis based on a non-linear model trained from the history of operations, which yields results that are difficult to detect using human rules.

In this sense, the obtained relations and networks of notarial operations allow the organization to enhance its usual analysis. Both the anomaly score and the detection of clusters provide guidance in the investigation of money laundering. In addition, the given graphs allow exploring the relations in a more visual way.

The Analysis Unit of this organization examined the operations detected using both solutions for possible reporting to the SEPBLAC, based on hints, in accordance to the provisions of Article 18 of Spanish Law 10/2010, of 28 April, on the prevention of money laundering and the financing of terrorism.

Proyectos relacionados