This research examines customer relationship management (CRM) systems using multi-criteria decision-making (MCDM) methods, with the intention of selecting the most suitable solution for small companies. The main goal of this research is to make a decision when choosing a CRM system by applying an objective approach. For this purpose, objective criteria were used, and an objective evaluation of the observed CRM systems was conducted. By using the MEREC (MEthod based on the Removal Effects of Criteria) method, the importance of the criteria was determined, while the CORASO (COmpromise Ranking from Alternative SOlutions) method was applied to rank the CRM systems. These methods were combined using a methodology into a hybrid approach. The results of this approach indicate that CRM systems with a higher degree of integration and automation achieved a higher rank, while systems with limited functionalities and longer implementation times received a lower ranking. This analysis confirms the importance of CRM systems in optimizing business processes, improving customer satisfaction, and enhancing marketing activities in companies. The results of the research can assist small companies in making decisions when selecting a CRM system. The contribution of this research is to provide efficient decision-making in the selection of a CRM system, thereby strengthening the companies' operations and improving their performance.
With the rapid advancement of modern robotics and artificial intelligence, intelligent picking robots have been widely adopted in agricultural production. Global path planning techniques have been applied to crop harvesting, such as oranges, apples, tea leaves, and tomatoes, yielding promising results. This study focuses on the path planning problem for a robotic arm used in premium tea leaf picking. Experimental simulations reveal that the Ant Colony Optimization (ACO) algorithm performs particularly well in solving small-scale Traveling Salesman Problems (TSP), as it can incrementally construct initial paths and, with properly tuned parameters, produce higher-quality solutions and achieve faster convergence compared to other algorithms. However, the traditional ACO algorithm tends to fall into local optima and suffers from slow convergence. To address these challenges, this paper proposes a dynamically optimized ACO algorithm that enhances the pheromone update rules and optimizes the $\alpha$ and $\beta$ parameters during the search process. These parameters are updated according to the optimization results, and a ranking factor is introduced to prevent the optimal picking path from being overlooked. The proposed method demonstrates superior performance over the traditional ACO algorithm in terms of path quality and convergence speed.
Accurate estimation of tree height is fundamental to sustainable forest management, particularly in regions such as Kumrat Valley, Pakistan, where Deodar Cedar (Cedrus deodara) serves as a vital ecological and economic resource. Conventional height estimation models often exhibit limitations in capturing the inherent complexity of forest ecosystems, where multiple environmental factors interact non-linearly. To address this challenge, a hybrid predictive framework integrating fuzzy inference systems (FIS) and multiple linear regression (MLR) has been developed to enhance the accuracy of height estimation. The FIS model incorporates key environmental and physiological parameters, including trunk diameter, soil quality, temperature, and rainfall, which are classified into fuzzy sets—low, medium, and high—corresponding to distinct growth rates (slow, normal, fast) and developmental stages (early, average, late). This classification enables a nuanced representation of environmental variability and tree growth dynamics. Complementarily, the MLR model quantifies the statistical relationships between these variables and tree height, yielding an R² value of 0.85, an adjusted R² of 0.64, and a statistically significant p-value of 0.04. The integration of fuzzy logic with regression analysis offers a robust, data-driven approach to height prediction, effectively addressing the uncertainties associated with environmental fluctuations. By leveraging both rule-based inference and quantitative modeling, this method provides valuable insights for precision forestry, contributing to the sustainable management and conservation of Deodar Cedar in Kumrat Valley.
Transformer-based language models have demonstrated remarkable success in few-shot text classification; however, their effectiveness is often constrained by challenges such as high intraclass diversity and interclass similarity, which hinder the extraction of discriminative features. To address these limitations, a novel framework, Adaptive Masking Bidirectional Encoder Representations from Transformers with Dynamic Weighted Prototype Module (AMBERT-DWPM), is introduced, incorporating adaptive masking and dynamic weighted prototypical learning to enhance feature representation and classification performance. The standard BERT architecture is refined by integrating an adaptive masking mechanism based on Layered Integrated Gradients (LIG), enabling the model to dynamically emphasize salient text segments and improve feature discrimination. Additionally, a DWPM is designed to assign adaptive weights to support samples, mitigating inaccuracies in prototype construction caused by intraclass variability. Extensive evaluations conducted on six publicly available benchmark datasets demonstrate the superiority of AMBERT-DWPM over existing few-shot classification approaches. Notably, under the 5-shot setting on the DBpedia14 dataset, an accuracy of 0.978±0.004 is achieved, highlighting significant advancements in feature discrimination and generalization capabilities. These findings suggest that AMBERT-DWPM provides an efficient and robust solution for few-shot text classification, particularly in scenarios characterized by limited and complex textual data.
Image segmentation plays a crucial role in medical imaging, remote sensing, and object detection. However, challenges persist due to uncertainty in region classification, sensitivity to noise, and discontinuities in object boundaries. To address these issues, a novel segmentation framework is proposed, integrating Complex Pythagorean Fuzzy Aggregation Operators (CPFAs) with a level-set-based optimization strategy to enhance both precision and adaptability. The proposed model leverages complex Pythagorean fuzzy membership functions, incorporating both magnitude and phase components, to effectively manage overlapping intensity distributions and classification uncertainty. Additionally, geometric constraints, including gradient and curvature-based regularization, are employed to refine boundary evolution, ensuring accurate edge delineation in noisy and complex imaging conditions. A key contribution of this work is the formulation of a complex fuzzy energy functional, which synergistically integrates fuzzy region classification, phase-aware boundary refinement, and geometric constraints to guide segmentation. The level-set method is utilized to iteratively minimize this functional, facilitating smooth transitions between segmented regions while preserving structural integrity. Experimental evaluations conducted across diverse imaging domains demonstrate the robustness and versatility of the proposed approach, highlighting its efficacy in medical image segmentation, remote sensing analysis, and object detection. The integration of complex fuzzy logic with geometric optimization not only enhances segmentation accuracy but also improves resilience to noise and irregular boundary structures, making this framework particularly suitable for applications requiring high-precision image analysis.