<p>Entity matching plays a critical role in data integration, decision-making, and interoperability across diverse datasets. Its effectiveness heavily depends on the design and integration of Similarity Measures (SMs), which quantify relationships between entities. However, no single SM consistently performs well across heterogeneous scenarios, necessitating an optimized combination for robust alignment. While Genetic Programming (GP) has been increasingly adopted for automating SM combination, it often generates complex and opaque feature sets, limiting interpretability and user trust. To address these challenges, this paper proposes a Tri-Objective Hybrid Genetic Programming (TOHGP) framework that optimizes SM combinations while ensuring matching accuracy, interpretability, and efficiency. Unlike traditional Single-Objective Evolutionary Algorithms (SOEAs) that optimize a single metric, TOHGP leverages a tri-objective evaluation strategy to effectively manage competing objectives. Additionally, an adaptive constant refinement mechanism dynamically adjusts constants in Pareto-optimal solutions, improving search performance without unnecessary computational overhead. Experimental experiments on OAE’s Conference datasets demonstrate that TOHGP consistently outperforms state-of-the-art methods, achieving higher-quality alignments with improved interpretability and reliability.</p>