Data-minding the Planet - iTrade Wildlife - software to detect illegal wildlife sales

Last modified by editors on 2023/05/22 20:09

WhatsOnSale.png
PIs: Dave Roberts and Michael Fischer
Funder: NERC

The illegal wildlife trade is fourth only to narcotics, human and counterfeiting in terms of transnational trafficking, and estimated to be worth $19-26.5billion pa. The UN recognised by consensus that environmental crime as an emerging form of transnational organised crime requiring a greater response by governments. The internet is a growing medium through which wildlife is being bought and sold due to the feeling of anonymity.

At the University of Kent, we have developed software, 'iTrade', that automatically identifies illegally traded items on websites with about 90% accuracy, capable of generating new reports every 10mins. Typically the identification of cybercrime of this form involves law enforcement officers simply going through thousands of web pages by hand. iTrade goes through all these pages automatically, weighting them as more or less likely to be illegal, and presenting the most likely pages in a feed so that officers can quickly click through these to assess investigation potential, saving officers a significant amount of time. For example it currently takes experienced officers a day to assess 400 items of ivory from the antiques section of eBay; 10-15% of these items have proved to be illegal elephant ivory. The software can acquire the 400 items and extract the 50-60 most likely items within ten minutes, which can be assessed in about hour on the part of the officer.

We will investigate the establishment of a business entity such as a non-profit organisation to deploy and develop services on advice gained through the Pathfinder. This will support our potential business model, which includes donations from third parties to support enforcement activities. Marketing research under the Pathfinder award indicated that our likely clients from the non-profit sector would be very reluctant to pay a private for-profit business for the same services, as would other likely donor institutions and private individuals. A non-profit places us on the same basis as the community within which we will operate, and will encourage volunteers and contributions of time from our colleagues. The organisation's income will contribute to improving and expanding services and products, supporting related research, and expanding our data-mining services to a broader range of applications. The University would maintain its interests through representation within the organisation.  webscrape.png

We will develop and deploy services based on present capabilities that supply feeds of webpages rich in illegal entries that can be deployed in a variety of forms, including RSS/Atom web feeds, email, and one or more web or tablet based applications, for use by public bodies with responsibilities to enforce including the National Wildlife Crime Unit of the Police and the UK Border Force, foreign law enforcement agencies, or external private parties who operate on behalf of law enforcement organisations.

In addition to facilitating direct use by public bodies, we will investigate the development of further vehicles for sponsorship of specific projects by relevant NGOs and foundations and develop a public online donation platform targeted at soliciting funds to promote specific wildlife enforcement initiatives. This will expand resources available to law enforcement for this purpose, increase the funds available to grow the organisation, and support additional research needed to address the core problem.

We will investigate the establishment of a consultancy division to directly work with a range of clients, advising on developing methods and workflows for detecting illegal activities on the internet. This will include expanding coverage to other kinds of illegal activity on the internet. For example, consultancy with groups such as the Ministry of Defence, Child Protection unit and English Heritage who have approached us, or the development of products that can evaluate other kinds of potentially illegal activity on the internet.

Normally the identification of cybercrime, such as the illegal wildlife trade, involves police, border force officers and independent consultants simply going through thousands of web pages by hand trying to recognise species or parts of for which trade is illegal (e.g. under the Convention on International Trade in Endangered Species). Our services platform, iTrade, collects and goes through all these pages, automatically weighting them as more or less likely to be illegal, saving as much as 90% of the time presently required. For example, it currently takes experienced officers a day to assess 400 items of ivory from the antiques section of eBay. This can be augmented by the software in a fraction of this time, creating a sample for evaluation that contains 90% probable illegal items rather than a typical 10% for a unfiltered set of web pages. Currently the software has been developed for eBay as a proof-of-concept. We are currently expanding to other websites, including Gumtree and Alibaba, based on a £10,000 grant from the University and a further £8,000 to develop image analysis software to identify Schreger lines in images that are indicative of ivory. We already have established partners, and therefore a customer base, including the Police (National Wildlife Crime Unit - NWCU), UK Border Force (UKBF) and the International Fund for Animal Welfare (IFAW) with whom the PI has been working with for over 10 years. Further discussions have taken place with the Dutch police force and the NGO Traffic regarding the service as part of the Pathfinder.

While we have a prototype product/service, a potential customer base, and a commercialisation plan, we lack the resources to employ a software engineer to take the product to a marketable stage and an established business structure and other resources to take the product forward. This proposal aims to advance our services to marketable quality and progress our business structure to support the deployment of our software services into the market place through further developing our IP and better matching it to the needs of our marketplace. As our business plan is currently based on relatively few organisations as regular customers, we have included resources to explore using social networks to increase capacity to raise funds. Ten specific objectives have been identified.

Technical Objectives

  1. Improve the User Experience with the workflow interface and support user customisation of the workflow interface.
  2. Support use cases proposed by NWCU and Dutch Police in the Pathfinder. 
  3. Refactor the services architecture to support client requests, new domains and workflows, improve adaptation to changes in data sources, and consider an online donations platform and social networking site for illegal wildlife sales.
  4. Improve the knowledge capture workflow and compile a 'plug in' collection of datasets trained by a range of domain experts to facilitate 'off the shelf' services for specific illegal wildlife domains. 
  5. Collect statistics to demonstrate and track productivity relative to current methods of detection.

Commercial Objectives

  1. Deploy easy to use and maintain applications and services supplying webpages rich in illegal entries for use by public bodies with responsibilities to enforce, and their collaborators (TO 1-5)
  2. Promote the development of sponsorship of projects from NGOs and social networking to solicit funds to promote wildlife enforcement initiatives (TO 2-5)
  3. Establish a consultancy division to directly work with clients, establishing new target domains (TO 4), and advising on and developing methods and workflows for detecting illegal activities (TO 3,5)
  4. Explore expansion of coverage to other kinds of illegal activity on the internet (TO 3,4)
  5. Better determine the form of business entity to invoke as we proceed, such as an annex to the University Enterprise Centre or a private non-profit entity to deploy, develop and sustain services (TO 3,4)
Tags: