Answers are grounded in this paper and cite their sources.
To address the novel timetabling challenges arising from China’s “3+1+2” college entrance examination reform, this paper proposes an adaptive beam search (ABS) framework to tackle the increased complexity of course scheduling under personalized education requirements. Specifically, the problem is decomposed into two interrelated stages: class formation and course timetabling. For each stage, corresponding mathematical optimization models are formulated, and tailored adaptive beam search algorithms are subsequently developed. In the class formation stage, a greedy strategy integrated with a state-aware beam search is employed to optimize student assignment and paired-class matching. In the timetabling stage, a multi-neighborhood search structure that explicitly incorporates paired-class constraints is introduced, together with a dynamic neighborhood selection mechanism to further improve the efficiency and effectiveness of the beam search process. Experimental results based on real-world data from a representative high school demonstrate that the proposed approach significantly outperforms mainstream algorithms as well as traditional manual scheduling methods in terms of both solution quality and computational efficiency. Moreover, ablation studies confirm the effectiveness of the proposed adaptive strategies. Overall, this work not only provides an efficient solution to highly constrained personalized timetabling problems, but also offers a scalable algorithmic paradigm applicable to a broader class of complex resource scheduling problems.
No takes yet. Share an insight, caveat, or question.
Song et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: