Loading

Menu

Revolutionizing
Mathematical Learning
with AI
Ontology Based Generative AI Model

Ontology-based OS : ONTOLOS

We develop task-specific AI Agents with educational precision and real-time responsiveness,

powered by a multi-agent system on an Ontology-based OS,

ensuring both educational sophistication and adaptive responses.

tech-mit-logo

Seomjae's engineering excellence is based on the ONTOLOS system, which is being researched in collaboration with MIT CSAIL. This system enables AI to break down and analyze traditional educational materials at the morpheme level, allowing for sophisticated personalized learning and feedback.

tech-mit-pictures
ONTOLOS

Structure

Layer 1 :
Data & Logic
Acquisition
This phase focuses on systematically acquiring essential educational data and formalizing the cognitive processes of domain experts.
bubble
Data
  • Collect educational data (e.g., lectures, quizzes) for content creation.
    • bubble
    Domain Logic
  • Capture expert decision-making to replicate in educational contexts.
    • Layer 2 :
    Ontologization
    This phase involves defining relationships among collected data and systematically connecting them to establish an interconnected data ecosystem.
    bubble
    Knowledge Graph
  • Organize concepts and relationships into a hierarchical structure.
    • bubble
    Purified Data
  • Convert raw data into a usable format with Knowledge Graph tags.
    • bubble
    Links
  • Connect data within the Data Hub for seamless interaction.
    • Layer 3 :
    Agentic AI Integration
    This phase focuses on designing and integrating AI agents capable of leveraging the ontology to deliver tailored solutions for diverse educational needs.
    bubble
    ConCreat Agent
  • AI that generates customized learning content.
    • bubble
    Tutoring Agent
  • Adaptive AI providing context-aware guidance.
    • bubble
    Learning Management Agent
  • AI that analyzes progress and offers recommendations.
    • ONTOLOS

    Core

    Ontologization
    Ontologization Layers
    Ontologization

    STEP 1

    Data Structuring

    Collaborating with domain experts and AI ontology specialists to design a Knowledge Graph that systematically defines relationships between mathematical concepts.

    STEP 2

    Algorithm Development

    Tagging essential conceptual information to content data to enhance usability.

    STEP 3

    Model Development

    Establishing connections between data using the tagged information through Linking.

    Agentic AI
    Integration

    STEP 1

    Analysis

    Designing an Agentic AI pipeline that mimics the cognitive processes of domain experts.

    STEP 2

    Pipeline Design

    Identifying and integrating AI models, algorithms, and technologies suitable for each stage of the pipeline.

    STEP 3

    Ontology Binding

    Binding Agentic AI with Ontology to ensure seamless real-time access to required information at each pipeline stage.

    Agentic AI Integration
    ONTOLOS

    Applications

    Application 1 :
    Tutoring Agent
    in Chatbot
    Teacher Persona Reflection:
    Adopts the tone and style extracted from the teacher's lecture data for a personalized teaching experience.
    Advanced Problem Solving:
    Solves problems using mathematical reasoning and provides explanations tailored to the student’s level.
    Context-Aware Q&A:
    Offers questions and answers suited to the user’s level and current status, enabling structured and meaningful conversations.
    Media Integration:
    Enhances student understanding through the use of images, videos, quizzes, and other interactive media.
    CHALK
    Application 2 :
    ConCreat Agent in
    Content Create System
    Content Analysis:
    Quickly identifies key features of problems, lectures, and solutions through analysis of video lectures, questions, and their explanations.
    Lecture Script and Voice Generation:
    Automatically generates lecture scripts (TST) and audio (TTS) aligned with the instructor's characteristics.
    Question and Solution Generation:
    Creates twin and similar problems, including detailed solutions, to enhance learning experiences.
    Video
    Audio
    Problem
    Solution
    Application 3 :
    Learning Management Agent
    in G-LMS
    Real-Time Feedback:
    Diagnoses student performance in real-time and provides personalized solutions such as problems, lectures, and quizzes.
    Real-Time Analysis:
    Offers instant insights and feedback based on abilities, weaknesses, and performance trends.
    Data Management:
    Proposes customized roadmaps to help students achieve their goals using their learning data.