Country Extraction Utility

Introduction

The Country extraction utility extracts country information out of a given address string, such a utility is useful in many application.

In order to successfully use the AML-TSD model, a country extraction utility must be constructed to obtain the opposite country of any wire examined.

Example:

The input file for this example will contain:

2221|389 Main St. Apt. 4C; Plainsboro, CT; 07060-38437 ENGLAND;3WD-384

The Output file for this example will contain:

2221|GBR|95|Country=England;State=;City=Plainsboro;Postal Code=07060-38437

Methodology

The approach for the country extraction process is a phased approached which relies on geographical information, in order to allow maximum accuracy in identification of country out of address information.
Geographical information consists of country names and their international codes (ISO2, ISO3), states names and codes, cities and postal codes.
The country identification is a part of the overall ETL process. The country extraction is a process done in batch format, for all wire source data.
The batch process expects and input file, it processes it and extract countries from the input data, scores the country match and outputs the country and the code into an output file.
The goal is to identify the opposite country of any wire examined. The utility’s configuration would allow the business users to fine tune the scoring done by the algorithm.

The country extractor detects the address in 3 phases:

• Parse preliminary input address string
• Divide string into regions and identify a country derived by each region.

Sum up the score voted per each country (per region) to conclude what is the country.

System requirements

In order to be able to achieve the relevant TPS the following hardware is required (or better):

• High Frequency Intel Xeon E5-2670 v2 (Ivy Bridge) Processors
• SSD-based instance storage for fast I/O performance
• 64-bit JVM

vCPU	Mem (GiB)	SSD Storage (GB)
2	7.5	1 x 32

**For More information please contact: [e-mail]