Scholarlink

<p>Frequent spikes in user activity during peak tax filing seasons often lead to system congestion, increased response time, and temporary service failures in digital tax filing platforms. These issues primarily arise because existing cloud infrastructures rely on reactive scaling mechanisms, which allocate resources only after demand has already increased. To address this limitation, this paper proposes an AI-based intelligent node allocation and workload tagging frame-work for large-scale digital tax filing systems.The framework formulates resource allocation as an optimization problem that minimizes latency, system failures, and resource overhead under dynamic workload constraints.The proposed archi-tecture integrates three interdependent components: a hybrid ARIMA–LSTM model for predictive workload forecasting, a reinforcement learning–based agent for dynamic resource scaling, and an income- and complexity-aware clustering mechanism for efficient workload distribution. The forecasting model captures both seasonal and non-linear demand patterns, achieving a prediction accuracy with a Mean Absolute Percentage Error (MAPE) of approximately 8.4%. Experimental evaluation under dynamic workload conditions demonstrates that the proposed framework reduces average job scheduling latency by 34.3% and minimizes system failures by up to 97.0% compared to traditional threshold-based cloud scaling policies. Further analysis through an ablation study shows that predictive scaling contributes most significantly to performance improve-ment, while clustering enhances system stability and resource efficiency. Overall, the results indicate that a unified approach combining demand fore-casting, adaptive resource provisioning, and intelligent workload management significantly improves the scalability and reliability of digital tax filing plat-forms. The proposed framework can be extended to other high-demand domains such as healthcare systems, financial services, and smart city infrastructures. The proposed architecture demonstrates practical applicability for large-scale government e-governance infrastructures operating under highly dynamic work-loads.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Smart AI-based node allocation for enhanced efficiency in digital tax filing networks

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月26日

查看全文
  • Eshaan Saha,
  • Tanya Mathur,
  • Pooja Gupta

摘要

Frequent spikes in user activity during peak tax filing seasons often lead to system congestion, increased response time, and temporary service failures in digital tax filing platforms. These issues primarily arise because existing cloud infrastructures rely on reactive scaling mechanisms, which allocate resources only after demand has already increased. To address this limitation, this paper proposes an AI-based intelligent node allocation and workload tagging frame-work for large-scale digital tax filing systems.The framework formulates resource allocation as an optimization problem that minimizes latency, system failures, and resource overhead under dynamic workload constraints.The proposed archi-tecture integrates three interdependent components: a hybrid ARIMA–LSTM model for predictive workload forecasting, a reinforcement learning–based agent for dynamic resource scaling, and an income- and complexity-aware clustering mechanism for efficient workload distribution. The forecasting model captures both seasonal and non-linear demand patterns, achieving a prediction accuracy with a Mean Absolute Percentage Error (MAPE) of approximately 8.4%. Experimental evaluation under dynamic workload conditions demonstrates that the proposed framework reduces average job scheduling latency by 34.3% and minimizes system failures by up to 97.0% compared to traditional threshold-based cloud scaling policies. Further analysis through an ablation study shows that predictive scaling contributes most significantly to performance improve-ment, while clustering enhances system stability and resource efficiency. Overall, the results indicate that a unified approach combining demand fore-casting, adaptive resource provisioning, and intelligent workload management significantly improves the scalability and reliability of digital tax filing plat-forms. The proposed framework can be extended to other high-demand domains such as healthcare systems, financial services, and smart city infrastructures. The proposed architecture demonstrates practical applicability for large-scale government e-governance infrastructures operating under highly dynamic work-loads.