Assessment: Cloze Test Generation

About

This system utilises multi-agent collaboration to automatically source and check content from the web and generate cloze (fill-in-the-blank) tests, publishing them with learning management systems (LMS). It employs a multi-agent workflow in which one agent handles real-time web searching via APIs, another formulates questions using LLMs and identifies significant entities to fill gaps, and quality is assured via automated validation metrics before deployment.

Features

  • Real-Time Sourcing: Automatically searches and retrieves content from the web for question generation.

  • Web Integration: Publishes assessments directly to learning management systems for real-time availability.

  • Automated Creation: Uses agents to formulate questions and select meaningful gaps for cloze items.

Latest Exemplars & Use Cases

Chinese Composition Assessment

Chinese Composition Assessment

This solution analyses photos of handwritten Chinese compositions to provide scores and feedback based on teacher-configurable rubrics. The technical core combines Optical Character Recognition (OCR) with LLMs for text extraction. A multi-agent system separates...

read more