Part 2 – Introducing K:blok – the digital solution to international trade and counterfeit challenges

Introduction

In February 2019, we (KOIOS Master Data) embarked on a successful year long research and development project focusing on “Using ISO 8000 Authoritative Identifiers and machine-readable data to address international trade and counterfeiting challenges”. This project was funded by Innovate UK, part of UK Research and Innovation. ISO 8000 is the international standard for data quality.

Part one of this article explains the challenges HMRC and the UK PLC face due to counterfeiting and misclassification when importing into the UK, and outlines a digital solution to solve those challenges. Upon which we won our Innovate UK grant.

This part of the article (part two) outlines the development progress made towards building a digital solution, how machine learning and natural language processing techniques were used during the year-long project and how the project can move forward.

K:blok – technology to traverse borders

To tackle the challenges outlined in part one, we developed a new software product, K:blok.

K:blok is a cloud application that allows importers to create a digital contract between the parties involved in the cross border movement of goods from the manufacturer to the importer/buyer. These parties can include: manufacturers, shippers, freighters, insurers and lawyers, amongst others.

The contract brings together, in a single source, various pieces of data that are required to successfully and efficiently import a product into the UK and data that is not currently captured in any software system:

• ISO 8000 compliant, machine readable, multilingual product descriptions produced by the manufacturer of the products;
• ISO 8000 compliant Authoritative Legal Entity Identifiers (ALEI’s) for each organisation that participates in the trade;
• Accurate commodity codes for each product, the quantity of products, serial numbers and anti-counterfeit information (only visible to the manufacturer, the buyer and HMRC) to help validate the authenticity of the product;
• Trade specific information required for insurance and accountability, for example: the trade incoterm;
• Licensing and trading information about the parties in the contract, for example: Economic Operators Registration and Identification (EORI) number;
• Information regarding the route the product is taking, for example: the port of import into the UK, port of export from the original country of export, vessel/aircraft numbers and locations of the change of custody of the consignments.

The contract is digital, machine readable, can be exchanged without loss of meaning and is suitable for interoperating with distributed ledger technology, like blockchain.

This data can be accessed and used by any of the participants of the contract and analysed by HMRC. All of this data is captured before the goods are moved which, in turn, provides an intelligence layer and pre-arrival data on goods for HMRC analytics, to enable resources to be targeted at consignments deemed high risk.

This single source of data also provides buyers with an audit trail for their purchased products, which begins with the original manufacturer which assists with the authentication of the product received and can form the basis of an efficient global trusted trader scheme.

Natural language processing will help avoid misclassification

As discussed in part one, misclassification leads to the UK losing billions in tax revenue. Misclassification is both intentional and unintentional. Reducing the unintentional misclassification could save the UK millions in tax revenue.

There is a fundamental flaw in the current process of tariff code assignment. The party that currently assigns the tariff code is not usually the manufacturer of the product. Therefore, the party does not have the technical knowledge to classify the product correctly. This party also rarely has a full description of the product and resorts to using a basic description from an invoice to assign the code.

Currently, HMRC provides an online lookup and email service to enable UK businesses to assign the correct tariff code. However, there are concerns that the service is not time efficient. This concern will only get worse as more companies may have to classify their goods once the UK leaves the European Union (EU).

Therefore, as part of our project, we worked with two students from the University of Southampton, studying Computer Science with Machine Learning, to create an additional application programming interface (API) that links with the government tariff code API and uses natural language processing techniques to score a similarity between an input product description and the potential mapping to the correct tariff code.

This is accessible by manufacturers using the KOIOS software to link their ISO 8000 compliant product specifications to the correct commodity code for trading with the UK.

Techniques such as term frequency-inverse document frequency (tf-idf) and K-means were integrated into this API. Support Vector Machine (SVM), Random Forest and a Deep Neural Network (2 layers) have also been explored to improve the accuracy of the algorithm.

The API successfully improves on the searching capabilities of the government online lookup service within the product areas explored in this project – which were bearings and couplings.

KOIOS are uniquely positioned to continue the development of digital solutions for the UK PLC

Our Innovate UK project provides a foundation to achieve more efficient, cost-effective, cross border trading and to reduce counterfeit activities. We believe that data standards, including ISO 8000 can play a huge part in digitising and automating this process further.

We are ideally suited and uniquely positioned to continue the research and development of both the K:blok platform and the machine learning tariff classifier.

We also believe there is an opportunity to digitise the outdated, human readable tariff classification into a digital classification, using the international standards ISO 22745 and ISO 29002. These data standards sit at the core of all of the products in the KOIOS Software Suite. A digital version of the tariff classification will improve the accuracy, speed and reliability of computer automation.

Join us in our vision

Our successful Innovate UK project was a step in the right direction to improving international trade and reducing counterfeiting. Brexit also provides a great opportunity for the UK to become a world leader in using technology across borders and to set the standard for countries to follow.

In the coming months, we will continue to engage with the UK Government/HMRC and continue to look for opportunities to fund our research and development.

If you think that you can add value to this project and would like to explore how we could collaborate then please get in touch at info@koiosmasterdata.com. 

Part 1 – The cost of counterfeit goods and misclassification to the UK

Introduction

In February 2019, we (KOIOS Master Data) embarked on a successful year-long research and development project focusing on “Using ISO 8000 Authoritative Identifiers and machine-readable data to address international trade and counterfeiting challenges”. This project was funded by Innovate UK, part of UK Research and Innovation. ISO 8000 is the international standard for data quality.

This part of the article (part one) explains the problem counterfeit goods and misclassification of products has on the UK PLC and the proposed solution which won us the Innovate UK Government grant.

A GBP 11 billion impact: and poor data exchange is the root of the problem

Counterfeit products and misclassification of products, when importing into the UK, cause major challenges for commercial organisations and the economy in the UK. These challenges increase a business’s exposure to risk, including consumer health, safety and well-being.

The impact of global counterfeiting on the UK economy is increasing. The Organisation for Economic Co-operation and Development (OECD) states that forgone sales for UK companies due to infringement of their intellectual property (IP) rights in global trade amounted to GBP 11 billion and at least 86,300 jobs were lost due to counterfeiting and piracy in 2019.

Protection from counterfeiting could save some organisations £000’s: for example, Greek customs seized 17,000 bearings, purporting to be from SKF, worth €1m in a single anti-counterfeiting operation.

When importing into the UK, importers are required to declare a commodity code for the products being imported. The commodity code is used to collect duty and VAT and dictates the restrictions and regulations, including the requirement for licensing, when importing or exporting the product.

Often the importer of the product does not have the technical knowledge to classify the product correctly. This, in combination with the complexity of the tariff code system currently adopted by the European Union (EU), and subsequently by the UK, causes many cases of misclassification. These cases are both intentional and unintentional.

Misclassification causes incorrect duty and VAT ratings to be applied to companies importing products, and also distorts trade statistics. Fraudulent misclassification leads to the UK losing billions in tax revenue.

Importers currently make customs declarations using the Customs Handling of Import and Export Freight (CHIEF) system, with some importers transitioning to the newer Customs Declaration Service (CDS).

The current importing process is not stringent enough and information is declared too late in the process. This results in a lack of transparency of the origin of products and a lack of quality data supporting the import and trade.

Therefore, Customs have the near impossible task of identifying and intercepting counterfeit or misclassified products. Customs activities increase spending by the UK Government on customs checks and delay trading activities.

International trade and counterfeit challenges: there is a digital solution

We believe that the challenges facing HMRC and the organisations that suffer from counterfeit goods can be solved with a stringent digital solution. A digital solution that captures:

• A quality description of the products in a consignment;
• The regulatory/licensing requirements on the products and the importer – for example: the commodity code of the product and the Economic Operators Registration and Identification (EORI) number; and
• The parties involved in the trade – for example: manufacturers, shippers, freighters, insurers and lawyers, amongst others,

in a timely manner (pre-arrival to the UK border). This assured, single source of data can then be used by all parties in the supply chain, including HMRC and border forces. HMRC will then be able to use this trusted data to better target resources on more risky consignments and the platform can be a requirement for inclusion in a trusted trader programme.

This digital solution can be taken further so that the importer using the platform can establish a purchase order with the seller.

We also believe that we can help to reduce the misclassification of products by:

1. Putting the responsibility of classifying the product on the manufacturer of the product, rather than the importer; and
2. Assisting the manufacturers with classifying the product by using ISO 8000 compliant, machine readable product specifications and machine learning techniques to search the current human readable tariff classification.

Without a digital data solution for the automating of tariff code assignment and the provenance of products, no significant improvements to the current state of play can be achieved.

The proposed solution would enable HMRC to: 

• reduce administration; 
• eliminate errors; 
• restrict growing levels of fraud in the digital economy 
• target resources effectively through collecting pre-arrival data on goods 

This proposal formed the fundamental basis of our successful Innovate UK grant application.

The next part of this article outlines the development progress made towards building a digital solution, how machine learning and natural language processing techniques were used during the year-long project and how the project can move forward.