4. - 10.50-11.20 - david mitzman - project manager - infocamere - introduction to bracco

Post on 18-Jul-2016

4 Views

Category:

Documents

1 Downloads

Preview:

Click to see full reader

DESCRIPTION

Presentation for Bracco.

TRANSCRIPT

Prevention of money-laundering and fiscal fraud: the BRACCO project

David Mitzman, InfoCamere, ITECRF 2010 - Budapest, Hungary

14 June, 2010

JLS/2007/ISEC/43130-CE-0220887/00-41

contents

• the BRACCO project• the BRACCO software system: what it does and how • the next steps

Intro to BRACCO

BRACCO is a 2-year project, from December 2008 to December 2010

It is co-financed by the EC Directorate General Justice, Liberty and Security

Partners: InfoCamere and Metaware (left the project in April)

Objectives of the project

The project contributes to the fight against fiscal crime by developing a search & investigation system to collect, aggregate, process, analyse and visualise information extracted from different public sources.

Such investigation is normally highly labour- and knowledge-intensive, requiring making sense out of a deluge of data.

Why should BRs get involved?

AML is not (necessarily*) a Business Register service, but it is an important area of collaboration between administrations.

* - some Registers are interested in monitoring their territories and all are interested in supporting thorough due diligence procedures.

the service context or market

The data sources include public registers such as the Business Register, the Insolvency and defaulters databases, yearly accounts statements, as well as newspaper articles.

Customer databases can also be integrated to provide additional relations for custom-tailored applications (Fiscal Police, Tax Authority, Stock Exchange Authority, Local Business Regulation Agencies)

the service context or market

A first (simple) product of this type is being used by the Fiscal Police and was introduced in 2009 to all the Italian Chambers of Commerce (BRs).

In 2010 it will be released as part of the standard Business Register access system.

A first (simple) product of this type is being used by the Fiscal Police and was introduced in 2009 into the Chamber of Commerce system.In 2010 it will be released as part of the standard Business Register access system.

the DeVisu navigator

Techniques in fiscal crime detection and prevention

• Integrate and aggregate different data sources• NLP for analysis of free-text information such as newspaper articles• Data Visualisation – “social networks”• Data Analytics - analyse networks for special features, to detect patterns or series of events• Notify relevant authorities of events

Relationships

Relations between companies and/or people:ownership, company roles, personal and professional relations or family ties, events such as transfer of shares, mergers, involvement in fiscal crimes and the related investigations

BRACCO software tools• analysis of free-text• data normalisation based on a model of the domain (rules and taxonomies governing the relationships - ontology) and based on a unique identification of subjects • an inference engine to further analyse the relations and draw conclusions (further relations) • graph analysis (“social network analysis”) and statistics on groups or clusters of companies • visual navigation of relations

three main layers of components

presentation layer

processing & storage

data sources

central knowledge

data analysis

BRACCO Ontology

data access

presentation layer

data sources

abstract and concrete knowledge

description of

real worldreal-world facts and

occurrences

data analysis

data access

presentation layer

data sources

structured and non-structured data

Taxonomies & rules

Relations graph

data analysis

DB adapters

presentation layer

Business

Register

Online Newspa

pers

Natural Language ProcessingPrivate-

client DBs

WWW spide

rs

analysing and enriching the knowledge

Taxonomies & rules

Relations graph

DB adapters

presentation layer

Business

Register

Online Newspape

rs

Natural Language ProcessingPrivate-

client DBs

WWW spiders

graph statistic

s warehou

se

inference engine

statistical

analysisgraph

analysis

outputs: navigation and reports

Taxonomies & rules

Relations graph (DB)

DB connectors

BRACCO Navigators(DeVisu, Redada, Soci-N)

Business Register

Online Newspape

rs

Natural Language ProcessingOther PA

DBs WWW

graph statistics warehous

e

inference engine

statistical analysis

graph analysis

Reports, dossier

REDADA navigator

investigating a person

inspecting relations and events

expanded node – F. Corona

inspecting a “sentimental relation”

Overview of crimes

Crimes tree

focus on location of activities

“Hotspots”

focus on recent activity

focus on recent activity

Business navigator

investigating InfoCamere

identifying InfoCamere group

What about Silvio Berlusconi?

... and his group?

finding hidden relationships

GOTCHA!!

Berlusconi “tree” [not the entire “web” of relations]

Indirect control – Spadaccini group

Group Dossier– Aerofotogrammetrica Nazionale

The next steps Further development of the data analysis techniques and statistics on groups

Application to nearby domains (Naples project on Public Tenders)

Validation with international shareholders data (Infogreffe); presentations and experiments (data exchange) with other BR’s

future steps data reliability measures

unique identification of subjects

new, customer-supplied relations

top related