Exam information
Public bilingual guides to exam formats, tasks, scoring and official preparation sources. These explain the exams; they are not the assessment Engine itself.
Visit Exams →Platform
AI Assessment Engine
AI-driven assessment infrastructure for English examinations. Transparent scoring, educator oversight, and reliable evaluation at scale.
System layers
AI Engine
Exam Modules
Study Environment
Overview
What the Engine Does
The Engine supports examination programmes and learning platforms with consistent, interpretable scoring and evaluation.
- Rubric-aligned scoring — Scores are tied to defined criteria and band descriptors so results map clearly to standards.
- Multi-skill evaluation — Supports speaking and writing assessment with criteria that reflect real examination design.
- Consistency and benchmarking — Calibrated models and quality controls help keep scores stable across time and cohorts.
- Confidence and reliability — Where the system can indicate reliability or confidence in a score, that information is surfaced for reviewers and users.
Design
Transparency by Design
We prioritise explainability so that educators and institutions can interpret and trust results.
- Score breakdowns — Subscores or dimension-level feedback where applicable, aligned to the same criteria used for the overall score.
- Rationale and criteria mapping — Outputs can be tied back to rubric dimensions so reviewers understand what drove the score.
- Audit trail — Systems are designed so that key decisions and overrides can be tracked for quality assurance and accountability.
Governance
Educator Oversight
The Engine is built for human-in-the-loop use. Automated scoring supports scale and consistency; educators retain the authority to review, override, or supplement results where policy or judgment requires it.
Our approach is grounded in real teaching and assessment experience. The design reflects how examiners and instructors actually evaluate performance, so that the system augments rather than replaces professional judgment.
Principles
Design Principles
-
Consistency
Scoring behaviour is stable and predictable across items and administrations, supported by calibration and quality controls.
-
Explainability
Scores and feedback can be traced to criteria and, where appropriate, to evidence in the response.
-
Responsible use
Systems are designed for oversight, override, and clear accountability rather than full automation without human review.
-
Practical utility
Outputs are actionable for instructors and learners, with reporting and workflows that fit real assessment practice.
Integration
Where It Connects
The Engine powers assessment and practice across EduZMS platforms. Two main entry points:
-
-
Learning environment
Practice and study tools that use the same assessment architecture for coherent feedback across learning and examination.
Visit Study →
Modules
Exam Information Portfolio
The public Exams site is the reference layer for current formats, task maps, scoring, preparation and official sources. It is separate from the assessment Engine described above.
-
IELTS
Academic and General Training guides covering skills, formats, band scoring and official preparation resources.
-
Cambridge
Live Pre A1–C2 qualification guides, including Young Learners and Schools variants.
-
TOEFL
Current January 2026 test order, adaptive task design, scoring and preparation guidance.
-
LanguageCert and China exams
Detailed LanguageCert Academic and LTE A1–C2 guides, plus ZhongKao, GaoKao, CET-4 and CET-6 references.