Network visualization for financial crime detection

1

Giuseppe (Beppe) Liotta

Network Visualization for Financial Crime Detection

UNIVERSITÀ DEGLI STUDI DI PERUGIA

Overview

•A crash introduction to Visual Analytics

•Visual Analytics for Crime Discovery

•A demo

2


Visual Analytics: Crash Intro

3

Visual Analytics: A “Dangerous” Zone?

4Drew Conway's Data Science Venn Diagram



Data mining by humanswho know math

Visual Analytics: A “Dangerous” Area?


Data mining by computersthat replace humans




Data mining by humansassisted by computers

Data mining by computersthat replace humans


A Conceptual Model for Visual Analytics

8

Data

Extraction

Data

ProcessingVisualization Interaction

Networked (Typically Big) Data Sets

•Data can often be modeled as a network (also called a graph) • they consist of entities (vertices) and relationships

(edges) connecting them

•Relational data are used in a large variety of application domains

9

Social Network Analytics

10

Bioinformatics

11

Sentiment (Political) Analysis

12


A Visual Analytics System for Crimes Investigations

13

Financial Crimes

• Financial crimes represent a

worldwide critical problem

• They are based on relevant volumes

of financial transactions to conceal the purpose, the source, or the destination of illegally gained money

• They weaken the integrity of worldwide financial markets

• They strengthen organized crimes that can mine the homeland security

Financial Intelligence Units

• To face this problem, most governments have created special investigation agencies, called Financial Intelligence Units

• The major challenge of financial subjects and of FIUs is to deal with the volume and the complexity of the collected data


FIU

INTELLIGENCE PRODUCTION

CYCLE

COLLECT

PROCESS

ANALYZE

DISSEMINATE


FIU


CYCLE

COLLECT

PROCESS

ANALYZE

DISSEMINATE

Persons, companies, banks, bank accounts, financial transactions…


FIU


CYCLE

COLLECT

PROCESS

ANALYZE

DISSEMINATE

Persons, companies, banks, bank accounts, financial transactions…

Financial Activity Networks (FANs)

Force directed e FAN

Visualizzazione dell'Informazione 21

vis4aia.mp4

vis4aia.mp4

User Evaluation

Q1 How much can VISFAN reduce the time and increase the productivity of the investigation processes in AIF?---------------------------------------------------------------------------------------------------

Q2 How much can VISFAN increase the effectiveness (i.e., the ability of identifying potential criminal activities) of the investigation processes in AIF?---------------------------------------------------------------------------------------------------

Q3 Within an investigation, how much do you benefit from the bottom-up exploration (incremental exploration) feature of VISFAN?---------------------------------------------------------------------------------------------------

Q4 Within an investigation, how much do you benefit from the top-down exploration (clustering) feature of VISFAN?

User Evaluation

Q5 How much do you benefit from the possibility of combining bottom-up and top-down exploration features of VISFAN in an effective way?

---------------------------------------------------------------------------------------------------

Q6 Within an investigation, how much do you benefit from imposing the geometric constraints of the network layouts in VISFAN?AIF?

---------------------------------------------------------------------------------------------------

Q7 How much are you satisfied with the interaction level of VISFAN?

---------------------------------------------------------------------------------------------------

Q8 Within an investigation, how much do you benefit from the set of centrality indexes of VISFAN?

---------------------------------------------------------------------------------------------------

Q9 Within an investigation, how much do you benefit from the tree-based saving system of VISFAN?

---------------------------------------------------------------------------------------------------

Q10 Within an investigation, how much do you benefit from the reporting functionality of VISFAN?

User Evaluation

1 = nothing

2 = not much

3 = enough

4 = a lot

Question Score

Q1 3.25

Q2 3.0

Q3 3.25

Q4 3.25

Q5 2.75

Q6 3.25

Q7 3.0

Q8 3.0

Q9 3.25

Q10 3.0


Thank You!!

25

Network visualization for financial crime detection

Technology

Transcript of Network visualization for financial crime detection