Researchers have introduced FSAM, a flexible similarity analysis model for Chinese medical short texts that achieves state-of-the-art accuracy while maintaining significantly lower computational overhead than competitive interactive models. When evaluated on two medical-domain benchmarks, the model achieved an accuracy of 86.45 percent on CHIP-STS and 92.46 percent on CQSJD.
Automated semantic similarity analysis of medical short texts serves as a crucial technical foundation for real-world healthcare applications, including clinical decision support and intelligent medical consultation systems.
Despite the importance of these applications, semantic similarity analysis remains challenging in existing models due to the prevalence of specialized terminology and complex contextual semantics. In addition, existing models struggle with an operational trade-off between classification accuracy and computational inference speed.
To address these challenges, the research team designed FSAM around an architectural component designated as the CTF module. This module integrates part-of-speech masking, a medical knowledge-enhanced similarity matrix, and an interactive attention mechanism to achieve multi-granularity semantic representation and alignment across input texts.
Ablation studies conducted by the researchers confirmed the substantial contribution of each component within the CTF module.
Beyond medical evaluations, the architecture demonstrated strong generalization capability across three public datasets from general and insurance domains, where it consistently delivered competitive performance. The authors noted that FSAM effectively balances accuracy and inference speed while generalizing well to out-of-domain datasets, highlighting its potential for deployment in healthcare natural language processing systems.
Subject of Research: Semantic similarity analysis of Chinese medical short texts
Article Title: FSAM: A Flexible Short Similarity Analysis Model for Chinese Medical Short Texts
Article References: Wang, H., Wu, H., Liu, K., Jiang, X., & Gao, Y. (2026). FSAM: A Flexible Short Similarity Analysis Model for Chinese Medical Short Texts. International Journal of Computational Intelligence Systems. https://doi.org/10.1007/s44196-026-01639-8
Image Credits: AI Generated
DOI: 10.1007/s44196-026-01639-8
Keywords: FSAM, medical short texts, semantic similarity analysis, Chinese natural language processing, CTF module, part-of-speech masking, medical knowledge-enhanced similarity matrix, interactive attention mechanism, CHIP-STS benchmark, CQSJD benchmark, clinical decision support, intelligent medical consultation, computational overhead, multi-granularity semantic representation, inference speed, healthcare NLP systems
News Source: Denise Maddox. (November 20, 2026). FSAM Model Improves Chinese Medical Text Similarity Analysis While Lowering Overhead. Scienmag.



