STFS_TF-IDF _HDSM: semantic and textual feature scaling using TF-IDF based sentiment analysis with Hybrid Deep Sequential Model

作     者：Kareem, Fadak R. Obaid, Ahmed J. 

作者机构：Faculty of Computer Science and Mathematics University of Kufa Najaf 54001 Iraq 

出 版 物：《International Journal of Information Technology (Singapore)》 (Int. J. Inf. Technol.)

年 卷 期：2025年

页      面：1-7页

主　　题：Attention mechanism Bi-LSTM Feature scaling Gated recurrent unit Sequential deep learning Suicidal sentiments TF-IDF 

摘      要：Sentiment analysis in healthcare is critical for the precise identification and assessment of suicidal intentions within clinical notes, enabling timely intervention and improving patient outcomes. The proposed study introduces a novel approach, STFS_TF-IDF_HDSM, where Term Frequency-Inverse Document Frequency (TF-IDF) is employed for Semantic and Textual Feature Scaling with the integration of Hybrid Deep Sequential Model (HDSM) to enhanced sentiment analysis. The variations among features are captured in sentimentsby utilising TF-IDF on the corpus that scales both semantic and textual features effectively. Textual context focuses on surface-level interpretation while semantics focuses on more complex and deeper understanding and reasoning about the text Sequential deep learning produces more precise and refined sentiment predictions in sentiment analysis by effectively capturing the context and emotional flow throughout text sequences. By integrating these Uni-Directional and Bi-Directional techniques with deep sequential processing to refine sentiment analysis for predicting potential suicide intentions achieved accurate results of 98.46% and precision 96.71% with minimum modal loss of 0.3121. This innovative framework demonstrates significant improvements in capturing refined sentiment patterns, offering robust performance across diverse textual datasets. © Bharati Vidyapeeth s Institute of Computer Applications and Management 2024.