Deceptive Domain Score

Features

Feature: Get the domain score
    As a domain analyst
    I want to know the deceptiveness score of a domain
    To decide whether or not it should be evaluated by a real person

Rules

• The domain will be compared to a list of keywords
• The input is a domain name
• Natural language must be matched (vísa == visa)
• The score is 100 if part of the domain that matches a keyword exactly and the keyword is a brand name
• The score is 90 if part of the domain matches a keyword exactly and the keyword is a common word
• The score is 80 if part of the domain is similar and the keyword is a brand name
• The score is 70 if part of the domain is similar and the keyword is a common word
• The score is 50 if the domain contains numbers
• The score is 0 if the domain is not in any of the categories listed above

Scenario: testing a deceptive domain
    Given the domain "visa-card-help.com"
    When I run the script on it
    Then it should result in a score of 100

Scenario: testing a domain that contains a number
    Given the domain "nfl-mega-league2019.com"
    When I run the script on it
    Then it should result in a score of 50

Code

Business Logic

Entities

• domain
• keyword

Public

• Domain
  • GetDeceptiveScore

Requirements

For the core business logic

• Go 1.12.x
• Remember to set export GO111MODULE=on;

For the AWS Serverless adapter

• Serverless Framework >=1.28.0
• An AWS account

Running tests

// In the root folder of the repository run
go test ./...

// To run tests showing the test coverage
go test ./... -cover

Building and deploying the AWS Serverless adapter

How to build

make build

How to deploy

make deploy or sls deploy

Deploying a single function

sls deploy -f clientget

TODO

• Create an architecture diagram
• Create a flow diagram for how the score is defined
• Remove tld (cleanup)
• Convert to lowercase (cleanup)
• Convert from punycode (cleanup)
• Different check for prefix/suffix (login, signin, verify...)
• Artificial intelligence (Machine learning) detection
• Create Go Benchmarks 💞
• Create Go Examples
• Run each algorithm in a separate Go Routine 😍 (based on Go Benchmarks, this is and slower option)