Entity–relation extraction constitutes a fundamental step in the construction of domain-specific knowledge graphs. In fault analysis of transmission systems, this task is complicated by extensive entity–relation overlap, nested structures, and strong semantic dependencies in technical texts. To address these challenges, an entity–relation joint extraction framework integrating reinforcement learning with a global pointer network (GPN) is developed (joint extraction model based on GPN and reinforcement learning, RL-BGPNet). A fault-oriented dataset is first established from helicopter transmission system maintenance manuals and related technical documents. Global semantic associations are then captured through a relation-aware attention mechanism, while parallel decoding is achieved using a GPN to accommodate overlapping and nested entities. The extraction of entity–relation triplets is further formulated as a multi-step decision process under a reinforcement learning paradigm, enabling coordinated optimization of entity recognition and relation classification and alleviating error accumulation caused by task interference. Experimental evaluations demonstrate that the proposed framework maintains stable performance under complex semantic conditions and exhibits satisfactory generalization, supporting its application to knowledge extraction and preliminary knowledge graph construction in the helicopter transmission system fault domain.
The strategic siting of a military airport constitutes a high-stakes planning problem characterized by complex trade-offs, long-term operational consequences, and pronounced uncertainty in expert judgment. In contrast to civilian airport planning, where economic efficiency and environmental externalities are typically prioritized, military airport location decisions are governed by additional requirements related to operational security, survivability, logistical resilience, and future capacity expansion. To address these challenges, a hybrid Multi-Criteria Decision-Making (MCDM) framework is proposed for the systematic evaluation and selection of military airport locations under uncertainty. Six core criteria and their associated sub-criteria, reflecting operational, strategic, technical, and infrastructural considerations, were identified through expert consultation and domain analysis. Criteria weights were derived using the Defining Interrelationships Between Ranked Criteria II (DIBR II) method and its Fuzzy, Grey, and Rough extensions, enabling the explicit modelling of vagueness, incompleteness, and ambiguity inherent in subjective assessments. Expert evaluations were aggregated using the Einstein Weighted Arithmetic Average (EWAA) operator, which accommodates heterogeneous levels of expertise and mitigates dominance bias. Alternative locations were subsequently ranked using the Weighted Aggregated Sum Product Assessment (WASPAS) method, allowing for flexible integration of additive and multiplicative aggregation schemes. The robustness of the obtained rankings was examined through a sensitivity analysis of the WASPAS aggregation parameter $\lambda$, confirming that variations in the aggregation structure do not alter the identification of the optimal and least-preferred alternatives. Furthermore, a comparative analysis with five established MCDM techniques revealed a high degree of rank correlation, thereby reinforcing the internal consistency and reliability of the proposed framework. The results demonstrate that the integration of uncertainty theories with advanced MCDM techniques provides a rigorous and adaptable decision-support tool for military infrastructure planning. Owing to its modular structure and methodological generality, the proposed framework can be readily adapted to diverse geographical settings, operational doctrines, and security environments, offering practical value for strategic decision-making in the defense sector.
Virtual communities function as large-scale knowledge interaction systems in which users jointly produce, exchange, and validate knowledge resources. However, not all interactions contribute positively to system performance. This study examined how different forms of value co-destruction behavior degrade knowledge interaction processes and user-level value outcomes in virtual communities. Drawing on survey data from 530 users of firm-hosted virtual communities, a structural equation modeling approach was employed to analyze the effects of five negative interaction behaviors—irresponsible behavior, knowledge hiding, avoidance, conflict, and negative information interaction—on three dimensions of user value: practical, entertainment, and social value. The results indicate that avoidance, conflict, and negative information interaction significantly reduce practical value by impairing knowledge accessibility and information reliability. Knowledge hiding, avoidance, and conflict significantly reduce entertainment and social value by weakening interaction quality and relational embeddedness. Interestingly, irresponsible behavior increases individual entertainment and social value while simultaneously posing systemic risks to collective knowledge quality. These findings suggest that value co-destruction is not merely a behavioral problem but a systemic phenomenon that degrades knowledge flow efficiency, information quality, and collaborative stability in digital knowledge ecosystems. The study contributes to knowledge engineering research by identifying key failure mechanisms in knowledge interaction systems and offers governance implications for designing resilient and sustainable online knowledge platforms.
In the context of technological development and digitalization of agriculture, mobile applications are playing an increasingly essential role in the management of small farms located in countries with fragmented agricultural structures. The aim of this research is to evaluate the most widely adopted mobile applications for monitoring and managing agricultural activities in areas with high agricultural potential such as Myzeqe, Korça, and Saranda in Albania. In order to achieve an impartial and sustainable assessment, multi-criteria decision-making (MCDM) methods integrated with fuzzy logic helped address the uncertainties and subjectivity in the evaluation process. The fuzzy CRiteria Importance through Intercriteria Correlation (CRITIC) method was employed to objectively determine the weights of the criteria based on the variability and contradiction between them. The fuzzy Combined Compromise Solution (CoCoSo) method was then adopted to rank the mobile applications. As revealed from the findings in this study, the most highly-rated criteria by experts, i.e., criterion C1-Ease of use and criterion C6-Integration with other technologies had the highest weight. The least rated criterion by experts was criterion C7-Technical support and training. AgriApp (A7) was the mobile application identified with the best performance. The contribution of this research lied in the building of a structured and objective framework to evaluate mobile technologies applied in agriculture, thus enabling more informed decisions for their adoption at the local and regional level.
The extent to which emotional perception shapes the acquisition, analysis, and presentation of knowledge within human–machine communicative interaction remains insufficiently understood. In this study, the principles of emotion artificial intelligentce (AI) (also referred to as affective computing) were integrated with trust as a socio-technical construct to investigate the mediating role of emotional expression in cognitive processing. A mixed-methods design was adopted, drawing on structured questionnaires and open-ended responses collected from 50 participants over a five-year period. Statistical modelling revealed that system quality significantly enhanced perceived ease of use when emotional signals were effectively encoded and decoded by both humans and machines. Trust was found to exert a positive influence on perceived usefulness, credibility, and user satisfaction, although it did not directly predict behavioural intention. In contrast, perceived ease of use demonstrated a strong positive association with intention in emotion-driven contexts, thereby rendering human–machine interaction more engaging, reliable, and trustworthy. These findings indicate that the tension between emotional and rational dimensions of higher cognitive processes within knowledge systems is shaped less by individual reluctance than by systemic and institutional determinants. The contribution of this work lies in the development of a conceptual framework for emotion-aware knowledge presentation, offering design implications for intelligent systems in education, public administration, business applications, and conversational AI. By demonstrating how emotion-aware mechanisms enhance both cognitive efficiency and affective engagement, the study advances understanding of human–machine cooperation and provides actionable guidance for the construction of more adaptive and trustworthy knowledge systems.
The macroeconomic performance of nations provides valuable insights into the knowledge economy and the governance structures that sustain its development. This study formalizes a framework for evaluating knowledge flows and innovation capacity through multi-criteria decision analysis (MCDA) using open World Bank data. The analysis employs the Logarithmic Decomposition of Criteria Importance (LODECI) method in conjunction with the Preference Selection Index (PSI) to determine objective weights, while the Weighted Euclidean Distance-Based Approach (WEDBA) is applied to rank the G7 countries and Türkiye in 2023. Knowledge flows, as represented by exports and foreign direct investment (FDI), serve as proxies for cross-border knowledge exchange, while inflation, unemployment, and economic growth are assessed within a reproducible, policy-driven framework. The weighting procedure assigns the greatest aggregate importance to inflation and the least to unemployment. The resulting rankings place the United States first, followed by Japan in second place, Türkiye fourth, and the United Kingdom last. The analysis further highlights how factors such as price stability, external openness, and investment dynamics shape national knowledge creation, diffusion, and organizational learning processes. By focusing on the utilization of open data, explicit knowledge representation, and transparent multi-criteria methodologies, the proposed framework strengthens digital knowledge infrastructures and facilitates actionable cross-country benchmarking. The findings have important policy implications, particularly in understanding how national macroeconomic variables influence innovation capacity. The framework is designed to be extensible, allowing for future adaptation to evaluate additional indicators, such as R&D intensity, high-tech export shares, and patenting activity. Furthermore, the approach is structured to support replication across various regions and timeframes, ensuring its broad applicability and scalability.
Accurate selective image segmentation continues to pose substantial challenges, particularly under conditions of noise interference, intensity inhomogeneity, and irregular object boundaries. To address these complexities, a novel framework is introduced that integrates fuzzy Einstein–Dombi (ED) operators with level set energy minimization, guided by marker-based initialization. The proposed approach departs from traditional intensity-driven models by jointly incorporating intensity, texture, and gradient-based features, thereby facilitating improved boundary delineation and enhanced regional homogeneity. A spatially adaptive regularization term has been embedded within the level set formulation to reinforce contour stability and robustness in the presence of artefacts and signal degradation. The fuzzy ED operators enable nuanced fusion of multiple features through non-linear aggregation, yielding a more expressive and resilient energy functional. In contrast to conventional segmentation schemes, the developed method achieves superior convergence and delineation accuracy, particularly within complex grayscale and noisy medical image datasets. Experimental validation has been conducted across a range of imaging conditions, with performance quantitatively assessed using established metrics, including segmentation accuracy (0.95), intersection over union (IoU: 0.89), and Dice similarity coefficient (DSC: 0.94). These results demonstrate statistically significant improvements over comparative models. Additionally, qualitative evaluations reveal enhanced contour fidelity and resistance to local intensity fluctuations. The methodological simplicity and computational efficiency of the framework render it highly suitable for real-time applications in medical imaging diagnostics, object detection, and related image analysis tasks. By offering a robust, interpretable, and generalizable solution, this work establishes a new reference point for selective image segmentation under non-ideal conditions, and paves the way for further exploration of fuzzy operator integration within variational segmentation paradigms.
For reducing uncertainty in data gathered from real-world scenarios, the picture fuzzy rough set (PFRS) framework is a reliable resource. This article presents new aggregation operators (AOs) based on the Schweizer-Sklar t-conorm (SS-TC) and Schweizer-Sklar t-norm (SS-TN). They present the PFRS framework with SS, which aims to handle the intricacies in contexts where decision-making is marked by ambiguity and uncertainty. In the context of Green Supply Chain Management (GSCM), where supply chain procedures incorporate sustainability considerations, this framework is especially pertinent. GSCM places a strong emphasis on minimizing environmental impacts by employing techniques such as effective resource management and sustainable sourcing. The adaptability and versatility required to assess and optimize these inexperienced practices are significantly improved with the aid of our expert PFRS framework. Businesses can keep operational efficiency and align their supply chain operations with environmental desires with the aid of using this framework. By considering both the blessings and disadvantages of environmental sustainability, using PFRS in GSCM enhances decision-making and promotes environmental sustainability. To handle picture fuzzy rough values (PFRVs), these operators include picture fuzzy rough weighted averaging (PFRSSWA) and picture fuzzy rough weighted geometric (PFRSSWG) operators. We investigate these recently created AOs' basic characteristics and use them to solve multi-attribute group decision-making (MAGDM) issues under the framework of picture fuzzy (PF) data. Our results demonstrate how the outcomes in SS-TN and SS-TC vary with varying parameter values. We also contrast these outcomes with the ones obtained from pre-existing AOs. In addition, we provide a graphic representation of all observations and findings to show how flexible and successful the suggested operators are at handling MAGDM problems.
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.
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.
Submit to IJKIS
Review for IJKIS
Propose a Special Issue 
Open Access
Journal Metrics
Journal Overview
Associated Journals