CaseLinker: An Open-Source System for Cross-Case Analysis of Internet Crimes Against Children Reports -- Technical Report & Initial Release
This paper introduces CaseLinker, an open-source system designed to facilitate cross-case analysis of child sexual exploitation and abuse reports by integrating hybrid information extraction, automated clustering, and interactive visualizations to overcome data fragmentation and reduce the emotional burden on investigators.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine a massive library where every book tells a heartbreaking story of a crime against a child. These books are scattered across hundreds of different towns, written in different handwriting, with some pages missing, others torn, and some written in code. The librarians (the investigators) trying to find patterns—like "Who is the same bad guy in three different towns?" or "What new trick are criminals using this year?"—are drowning. They have to read thousands of these disturbing stories one by one, which is not only incredibly slow but also causes them deep emotional pain, like watching a horror movie on repeat for years.
CaseLinker is a new, open-source tool designed to be a "super-librarian" that helps these investigators without making them read every single page of every book.
Here is how it works, broken down into simple concepts:
1. The Problem: A Mountain of Disorganized Files
Right now, if an investigator wants to know if a criminal is active in multiple states, they have to manually read through thousands of PDF reports.
- The Analogy: Imagine trying to find a specific type of red car in a parking lot the size of a city, but the cars are parked in different lots, some are covered in tarps, and the descriptions are written in different languages. Doing this by hand is impossible.
- The Human Cost: The people doing this work are constantly exposed to the worst parts of human behavior. It's like asking a firefighter to sit in a burning building for 12 hours a day, every day. They get "burned out" (traumatized) and can't think clearly anymore.
2. The Solution: CaseLinker (The Smart Sorter)
CaseLinker is a computer program that acts like a high-speed, super-organized filing clerk that never gets tired and never gets emotionally hurt.
Step 1: Reading the Messy Notes (Ingestion)
The system takes the messy PDF reports (which look like unorganized notes) and turns them into clean, digital text.
- Analogy: It's like a scanner that not only copies a handwritten letter but also fixes the spelling mistakes and organizes the paragraphs so a computer can read them.
Step 2: The "Detective" (Processing)
Instead of using complex, mysterious AI that acts like a "black box" (where you don't know how it thinks), CaseLinker uses clear, rule-based searching.
- Analogy: Imagine a detective who has a checklist. If the text says "Facebook," the detective checks the "Social Media" box. If it says "baby," they check the "Infant Victim" box.
- Why this matters: Because the rules are simple and visible, a judge in court can look at the computer's work and say, "Yes, I see exactly why it flagged this." It's transparent and trustworthy.
Step 3: The "Grouping" (Clustering)
Once the data is organized, CaseLinker starts grouping similar cases together.
- Analogy: Think of it like a music app that says, "Hey, these 50 songs all sound like '90s Rock,' so let's put them in one playlist."
- CaseLinker creates playlists like:
- "The Online Predators": Cases involving chat rooms and social media.
- "The Severe Cases": Cases involving the youngest victims or the worst crimes.
- "The Repeat Offenders": Cases where the criminal has been caught before.
- It uses a math formula (Jaccard similarity) to figure out how much two cases look alike, even if they happened in different states.
Step 4: The "Priority Alarm" (Triage)
Not all cases are the same. Some need immediate attention; others can wait. CaseLinker gives every case a "danger score" from 1 to 10.
- Analogy: It's like a hospital triage nurse. If a patient is bleeding out (high severity, infant victim), they get a red tag and seen immediately. If it's a minor cut, they get a green tag.
- This helps investigators focus their energy on the most dangerous criminals first, rather than wasting time on low-priority paperwork.
Step 5: The "Safe Dashboard" (Visualization)
This is the most important part for the human investigators. Instead of showing them a wall of text or graphic images, CaseLinker shows them charts, graphs, and color-coded bars.
- Analogy: Instead of forcing a doctor to look at a raw, bloody wound to diagnose a disease, the machine gives them an X-ray and a temperature chart.
- The investigator can click on a bar chart showing "Facebook" and instantly see a list of cases, with the scary words highlighted in yellow so they don't have to read the whole story. This protects their mental health while still giving them the answers they need.
Why This is a Big Deal
- It's Free and Open: Anyone can download it. It doesn't cost millions of dollars like the fancy tools big police departments use.
- It Protects the Helpers: By doing the heavy lifting of reading and sorting, it saves investigators from the trauma of reading thousands of horrific stories.
- It Finds Hidden Patterns: It can spot that a criminal is using a specific app in three different states, something a human reading one report at a time would never notice.
- It's Honest: Because it uses simple rules instead of "magic AI," the results can be used in court to put criminals away.
In Summary
CaseLinker is a tool that turns a chaotic, traumatic pile of paperwork into a clear, organized map. It helps investigators catch bad guys faster, protects the mental health of the people doing the work, and ensures that justice isn't delayed because someone was too overwhelmed to read the next file. It's technology serving humanity by saying, "You do the thinking; we'll do the reading."
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.