fraud-detection

This is an application intended to persist data about users and transactions in order to detect fraud in the context of e-commerce. With this purpose, it provides a REST API to inject the data and check if a transaction is a fraud attempt.

It also provides a Kafka consumer and a RabbitMQ listener to listen to user data injection commands.

It uses Spring Data Neo4j to model a graph database. This way of connecting entities is more natural and efficient for the purposes of the application.

Plus, it offers the possibility to integrate CAMARA Project fraud detection APIs. For now, the Device Location Verification API is integrated through MultiApi Maven Plugin. The real connection with real telco companies depend on these considerations. Most telco companies require registration or charge for the use of their APIs. It has been intended that the final considerations (like consent management and the relation with the API providers) about the use of these APIs is delegated to the potential users of this application. A revision on the preparation of these matters for final users is pending.

Technologies

• Java 17
• Spring Boot
• Neo4j
• Junit5
• Maven
• Kafka
• RabbitMQ

How to run on your host

1. Ensure Docker and Docker Compose are installed on your system.

2. For the tests to pass, you will need an instance of Neo4j running. You can start it with Docker by running:

   docker run --publish=27474:7474 --publish=27687:7687 --volume=$HOME/test/neo4j/data:/data --env NEO4J_AUTH=neo4j/secretgraph neo4j

3. Create the .jar application by running the following command in this repository root:

   mvn clean package -P kafka,rabbitmq

   The activation of "kafka" and "rabbitmq" profiles ensures the creation of the .jar with the necessary dependencies to run the Kafka consumer and the RabbitMQ listener, repectively.

4. Build the fraud-detection-rabbitmq-kafka image by running the following command in this repository root:

   docker build -t fraud-detection-rabbitmq-kafka .

5. Start the services with:

   docker-compose up

This will start all required services, and the fraud-detection REST endpoint of the application will be available at port http://localhost:8080.

You can access Neo4j at http://localhost:7474, RabbitMQ Management UI at http://localhost:15672, and use the application's endpoints to test the fraud detection system.

REST Endpoints

/users

POST /inject (registers a User in the database)

Parameters

name	type	data type	description
injectUserCommand	required, body	object (JSON)	Object containing all the necessary information about a user to be stored (InjectUserCommand.class)

Responses

http code	content-type	response
200	text/plain;charset=UTF-8	User injected successfully
400	application/json	a field must not be null, empty, or invalid (e.g. email format)

/devices

POST /inject (registers a Device in the database, related to a User)

Parameters

name	type	data type	description
injectDeviceCommand	required, body	object (JSON)	Object containing all the necessary information about a Device to be stored (InjectDeviceCommand.class)

Responses

http code	content-type	response
200	text/plain;charset=UTF-8	Device injected successfully
400	application/json	a field must not be null, empty, or invalid (e.g. email format)
404	application/json	the User of the Device does not exist in database

/locations

POST /inject (registers a Location in the database, related to a Device)

Parameters

name	type	data type	description
injectLocationCommand	required, body	object (JSON)	Object containing all the necessary information about a Location to be stored (InjectLocationCommand.class)

Responses

http code	content-type	response
200	text/plain;charset=UTF-8	Location injected successfully
400	application/json	a field must not be null nor empty
404	application/json	the Device indicated does not exist in database

/transactions

POST /inject (registers a Transaction in the database, related to a User, a Device and a Location)

Parameters

name	type	data type	description
registerTransactionCommand	required, body	object (JSON)	Object containing all the necessary information about a Transaction to be stored (RegisterTransactionCommand.class)

Responses

http code	content-type	response
200	application/json	the Transaction successfully registered, including a field indicating if it has been declared as a fraud attempt
400	application/json	a field must not be null nor empty
404	application/json	the Device, User or Location indicated does not exist in database