Magnet Forensics has a new white paper available that discusses why and how contextual content analysis works, based on a description of what child luring is (and isn’t) and what nuances to expect.
The white paper also covers how machine learning, such as the model introduced with Magnet.AI (our industry-leading technology that analyzes conversations to recover potential child luring content for examiners to consider), overcomes the above challenges using data science.
Download our white paper now!
New white paper also describes:
- Why “accuracy” isn’t enough of a metric, and how to balance it with a second metric, “precision.”
- How to deploy contextual content analysis within a typical case workflow.
Like triaging images to focus on those with the highest likelihood of containing contraband, investigators of child exploitation need a way to triage content for the messages that contain illicit conversations. Magnet.AI (read more about Magnet.AI here: Introducing Magnet.AI: Putting Machine Learning to Work for Forensics) offers that capability, identifying whether those devices have been used to lure, or groom, children for sexual activity.
Magnet.AI was developed in response to the difficulties that investigators and forensics examiners who work child exploitation cases can have establishing that a conversation taking place is “luring” or “grooming”—a conversation that builds a quasi-relationship for the ultimate purpose of abusing a child.
Those challenges include a number of factors:
- The sheer number of conversations. Suspects may pursue many more than one victim over the course of dozens or even hundreds of conversations each, targeting victims via chat apps, chat functions embedded within games, instant or text messaging apps, and more across computers, smartphones, gaming consoles, and tablets. This raises the risk of human error—missing a key piece of evidence.
- The conversations themselves can appear innocent, especially between family members, or between the child and an adult they trust (such as a teacher or coach). Conversely, illicit-appearing conversations can be taking place between consenting adults.
- The nature of conversations can be very different according to intent, as well. Child predators’ goals, timing, and methods are not necessarily the same as human traffickers’ or opportunistic child abusers.
- Investigators and examiners are under intense pressure by stakeholders and the public in almost every case, and in those involving children especially. Getting quick insight into evidence and being able to make informed, credible fast decisions can be crucial.
Identifying malicious intent, therefore, is finding the proverbial needle in the haystack—without knowing which haystacks to start with. Moreover, language itself is unstructured. Slang, regional dialect or references, and terminology used among different groups, as well as internet shorthand, can all factor into whether an investigator correctly identifies an illicit conversation.