Skip to main content
Req. Number
< All Topics
Print

Req 2.42: AI-Based Risk Detection Engine

📋 Requirement Details

Req 2.42: AI-Based Risk Detection Engine
1. Detect indicators of self-harm, suicide, bullying, violence, drug use, and exposure to inappropriate content using AI and contextual-language analysis.
2. Support detection in English, Tagalog, commonly used Filipino colloquialisms, and other terms that may subsequently be supplied by DepEd.

⚙️ Technical Demonstration

IMTLazarus incorporates an advanced, real-time Contextual Threat Detection & Behavioral Intelligence Engine that runs directly at the endpoint level to identify safety risks before they escalate.
The system features a native natural language processing (NLP) and contextual analysis engine that scans text inputs, search queries, chat applications, and web pages for high-risk behavioral indicators. It evaluates semantic context rather than just static keywords to detect explicit signs or patterns of self-harm, suicidal ideation, cyber-bullying, explicit violence, illegal drug use, and exposure to adult or inappropriate content. Upon detection, the platform logs the context, applies defensive content mitigations (such as real-time image blurring), and triggers silent administrative alerts.
To ensure comprehensive localized safety coverage within the Philippine educational environment, the language processing engine supports native multi-dialect parsing:

English and Core Directories: Standard processing matrices cover comprehensive global definitions for safety risks.

Tagalog and Local Colloquialisms: The system features localized keyword filtering and NLP directories configured specifically for Tagalog, contemporary Filipino internet slang, and colloquial variations used by students in online communication.

Extensible Custom Libraries: The platform features an open, dynamic dictionary framework that allows DepEd Central Office administrators to instantly push new, localized terms, emerging slang words, or regional dialect variants over-the-air to the entire fleet's detection baseline.

🖼️ Visual Evidence

Table of Contents