This paper introduces an Inductive Cognitive Diagnosis Model (ICDM) for fast inference of new students' mastery levels in web-based online intelligent education systems (WOIESs). The ICDM addresses the challenge of efficient cognitive diagnosis in open learning environments where a large number of new students continuously register and complete exercises. Unlike existing transductive cognitive diagnosis methods that rely on student-specific embeddings, ICDM employs a novel student-centered graph (SCG) to derive inductive mastery levels as the aggregated outcomes of students' neighbors. This approach shifts the task from updating student-specific embeddings to finding suitable representations for different node types in the SCG, eliminating the need for retraining. The ICDM consists of a construction-aggregation-generation-transformation process to learn final representations of students, exercises, and concepts. Extensive experiments on real-world datasets demonstrate that ICDM is significantly faster than transductive methods while maintaining competitive inference performance for new students. The model also shows improved interpretability and consistency in inferred mastery levels, making it suitable for downstream tasks such as learning item recommendation and computerized adaptive testing.This paper introduces an Inductive Cognitive Diagnosis Model (ICDM) for fast inference of new students' mastery levels in web-based online intelligent education systems (WOIESs). The ICDM addresses the challenge of efficient cognitive diagnosis in open learning environments where a large number of new students continuously register and complete exercises. Unlike existing transductive cognitive diagnosis methods that rely on student-specific embeddings, ICDM employs a novel student-centered graph (SCG) to derive inductive mastery levels as the aggregated outcomes of students' neighbors. This approach shifts the task from updating student-specific embeddings to finding suitable representations for different node types in the SCG, eliminating the need for retraining. The ICDM consists of a construction-aggregation-generation-transformation process to learn final representations of students, exercises, and concepts. Extensive experiments on real-world datasets demonstrate that ICDM is significantly faster than transductive methods while maintaining competitive inference performance for new students. The model also shows improved interpretability and consistency in inferred mastery levels, making it suitable for downstream tasks such as learning item recommendation and computerized adaptive testing.