On 15 March, INHOPE is launching a common schema for the classification of child sexual abuse and exploitation material. This schema was created as a part of the Global Standard Project and aims to define and normalize terminology that is utilized by Industry, Hotlines, Non-Government Organizations (NGOs), and Law Enforcement regarding Child Sexual Abuse Material (CSAM).

What is the Global Standard project?

Funded by End Violence Against Children, the goal of the Global Standard Project is to create a common schema for the classification of child sexual exploitation and material schemas to, in turn, facilitate better collaboration for identifying illegal material, to aid in the creation of annotated datasets for the training of automated CSAM detectors, and produce more relevant reporting, all for the purpose of better victim identification capabilities and more effective processing of CSAM by hotline analysts, law enforcement officers and technology industry professionals. Countries have greatly differing definitions of illegal material, and organisations that assess child sexual abuse and exploitation material categorise and hash content based on their national legislation. Meaningful International cooperation has been hampered by the lack of a common, globally compatible system that would make media categorised to criteria specific to one country understandable and operationable by another.

The schema was created to overcome these barriers – with a core goal being normalising the language and terminology used in classficiation to promote comparability between jurisdictions.

How was the schema created?

The foundation for the schema was developed from a diverse array of existing ontologies and categorisation schemas, including but not limited to the International Categorisation Collaborative (IC3) developed by Argos at Queensland Police Service Australia and INTERPOL DevOPS, the US Industry Standard, the NCMEC categorisation system, the Intelligrade system developed by the Internet Watch Foundation, and Project VIC. For the past 9 months, an international working group of experts from law enforcement agencies, the largest INHOPE hotlines, NGOs in the field, and technology partners have used the foundations of this body of work to develop the ontology.

The schema was developed to meet the needs of all stakeholder groups involved in the categorisation or implementation of CSEM/CSAM-based datasets for a diverse range of purposes. This is made possible through a standardisation of the terminology relevant to determining the illegality of Child Exploitative content, whether the information is used for the minimum purpose of assessing illegality, to the more granular content-based and context-based use cases for machine learning and training. This ontology forms the basis of a system to aid in the determination of the illegality of media for a given jurisdiction regardless of categorisation schema and legislation.

What does the Universal Classification Schema aim to do?

• Realign Machine Learning efforts for automated CSAM detection
• Normalize language and terminology to facilitate better compatibility between systems and hash sharing
• Improve collaboration between sectors
• Reduce exposure on investigators
• Improve victim identification capacity
  
  

The Universal Classification Schema is being made available upon request to select relevant stakeholders for testing and mapping to their internal systems & schemas. If you’re interested in receiving more information about the schema send a request to globalstandard@inhope.org.