Institutional Repository of Academic Research University of Macedonia
Not a member yet
2215 research outputs found
Sort by
Temporal Action Analysis in Metaheuristics: A Machine Learning Approach
No full textThis study explores the use of Autoregressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) machine learning models in metaheuristic algorithms, with a focus on a modified General Variable Neighborhood Search (GVNS) for the Capacitated Vehicle Routing Problem (CVRP). We analyze the historical chain of actions in GVNS to demonstrate the predictive potential of these models for guiding future heuristic applications or parameter settings in metaheuristics such as Genetic Algorithms (GA) or Simulated Annealing (SA). This “optimizing the optimizer” approach reveals that, the history of actions in metaheuristics provides valuable insights for predicting and enhancing heuristic selections. Our preliminary findings suggest that machine learning models, using historical data, offer a pathway to more intelligent and data-driven optimization strategies in complex scenarios, marking a significant advancement in the field of combinatorial optimization.1475336537
A Comparative Study of Sentiment Classification Models for Greek Reviews
No full textIn recent years, people have expressed their opinions and sentiments about products, services, and other issues on social media platforms and review websites. These sentiments are typically classified as either positive or negative based on their text content. Research interest in sentiment analysis for text reviews written in Greek is limited compared to that in English. Existing studies conducted for the Greek language have focused more on posts collected from social media platforms rather than on consumer reviews from e-commerce websites and have primarily used traditional machine learning (ML) methods, with little to no work utilizing advanced methods like neural networks, transfer learning, and large language models. This study addresses this gap by testing the hypothesis that modern methods for sentiment classification, including artificial neural networks (ANNs), transfer learning (TL), and large language models (LLMs), perform better than traditional ML models in analyzing a Greek consumer review dataset. Several classification methods, namely, ML, ANNs, TL, and LLMs, were evaluated and compared using performance metrics on a large collection of Greek product reviews. The empirical findings showed that the GreekBERT and GPT-4 models perform significantly better than traditional ML classifiers, with BERT achieving an accuracy of 96% and GPT-4 reaching 95%, while ANNs showed similar performance to ML models. This study confirms the hypothesis, with the BERT model achieving the highest classification accuracy.8910
A Link-Quality Anomaly Detection Framework for Software-Defined Wireless Mesh Networks
No full textSoftware-defined wireless mesh networks are being increasingly deployed in diverse settings, such as smart cities and public Wi-Fi access infrastructures. The signal propagation and interference issues that typically characterize these environments can be handled by employing SDN controller mechanisms, effectively monitoring link quality and triggering appropriate mitigation strategies, such as adjusting link and/or routing protocols. In this paper, we propose an unsupervised machine learning (ML) online framework for link quality detection consisting of: 1) improved preprocessing clustering algorithm, based on elastic similarity measures, to efficiently characterize wireless links in terms of reliability, and 2) a novel change point (CP) detector for the real-time identification of anomalies in the quality of selected links, which minimizes the overestimation error through the incorporation of a rank-based test and a recursive max-type procedure. In this sense, considering the communication constraints of such environments, our approach minimizes the detection overhead and the inaccurate decisions caused by overestimation. The proposed detector is validated, both on its individual components and as an overall mechanism, against synthetic but also real data traces; the latter being extracted from real wireless mesh network deployments.249551
Clusterslice: Slicing resources for zero-touch Kubernetes-based experimentation
No full textClusterSlice is an open-source solution for automated Kubernetes-centered experimentation. It introduces well-designed abstractions that reduce experimentation complexity with improved reliability and reproducibility. Its main capabilities are: (i) automated declarative operation, e.g., through declarative specifications of experimentation slices; (ii) infrastructure-as-a-service (i.e., the utilization of heterogeneous physical and virtual resources), platform-as-a-service (e.g., multiple composable Kubernetes flavors and network plugins), and application-as-a-service (e.g., plug-and-play application features) capabilities; (iii) multi-cluster and multi-domain support, i.e., inter-cluster operation over multiple open testbeds, virtualization systems and domains; and (iv) experimentation automation, e.g., support of automated experiments of various research topics, including network plugin performance assessments and anomaly detection workflows. Here, we provide the basic architectural attributes of ClusterSlice and proof-of-concept results highlighting its above capabilities.16111
Selection of Academic Staff Based on a Hybrid Multi-criteria Decision Method Under Neutrosophic Environment
No full textMulti-criteria decision-making (MCDM) is a decision-making process that involves assessing and selecting the best alternative from a group of options based on various criteria or qualities. In this research work, we propose and elucidate the theory of neutrosophic logic, which is unique in its approach to evaluating candidates’ performance in a manner that takes into account significant elements and criteria that are essential for the overall process when dealing with unclear, inaccurate, or incomplete data. We propose a novel hybrid integrated MCDM methodology based upon neutrosophic Delphi (N-Delphi) and neutrosophic AHP (N-AHP) methods, which takes into consideration the importance of each decision-maker and their preferences per evaluation criterion. A new MAXMIN threshold value technique treats the criteria under consideration as the decision alternatives and their score functions as their payoff values, thus reducing unnecessary resources by eliminating unimportant criteria during the personnel selection process.5
The nonlinear relationship between employee stock ownership plans and firm performance: Evidence from China
No full textTwo opposing forces (positive and negative) have been theoretically expounded to explain the relationship between employee stock ownership plans (ESOPs) and firm performance. However, the direction of this relationship remains a puzzle, especially in countries where state ownership is significant. This paper draws on the “Guidance on the pilot implementation of employee stock ownership plan (ESOP) by listed companies” issued by the China Securities Regulatory Commission (CSRC) to assess the moderating effect of state ownership on the relationship between ESOPs and firm performance. Using employee and executive stock ownership data from 620 Chinese listed companies between 2014 and 2020, I find an inverted U-shaped relationship between employee stock ownership and firm performance. However, the inverted U-shaped relationship holds only for non-state-owned Chinese firms. Finally, a U-shaped relationship between executive ownership and company performance has been found.17311447
A Methodology for Developing & Assessing CTI Quality Metrics
No full textSince its first steps in the cybersecurity field, Cyber Threat Intelligence (CTI) has gained recognition and increased its importance in the daily operations of cybersecurity teams. However, the many forms of CTI exchanged, the vast amount of CTI products, and the plurality of the sources have raised doubts about the CTI quality. This paper discusses the problem of CTI quality, focusing on the quality factors that better evaluate the products of CTI and how we measure them. Consequently, we propose a methodology for developing and assessing CTI quality metrics and demonstrate the application of this methodology by developing the relevance (RE) and weighted completeness (WC) metrics for unstructured and structured CTI products, respectively. We created two sets of structured and unstructured CTI data for this demonstration, utilizing them as benchmark datasets for estimating RE and WC.The proposed methodology introduces a systematic approach for developing and assessing quantitative CTI quality metrics for evaluating CTI data and CTI sources.126225623
A Semi-Automated Approach for Resolving Data-Driven Architecture Mismatches
No full textIn contemporary software development, there is a need for delivering solutions that require the integration of multiple software systems, each one relying on different architec-tural decisions. For instance, e-shop solutions must communi-cate with the ERP solution that the company possesses to handle prices, products, and stock. However, such an integration is not always a trivial issue since interoperability problems might arise. A root cause for such interoperability issues is architecture mismatches: e.g., caused by heterogeneity on how data are stored and are expected to be exchanged in the two systems. In-teroperability problems can cause delays to the development, require extended communication with different teams, and usually adds complexity to the system. In this paper, we propose a semi-automated AI-based approach and a middleware software solution ('a connector') to aid software engineers in 'connecting' applications with heterogeneous data storing schemas. We have validated our approach and tool with a company that connects ERP systems with e-shops, through a qualitative study.17Proceedings of the 2024 IEEE 21st International Conference on Software Architecture Companion (ICSA-C
Digital Strategy and Change in Public Services and Enterprises: The Case of IRIDA Document Management Information System
No full textDigital transformation is recognized as a phenomenon that has significantly changed how organizations operate. The emergence of digital technologies in the public sector has presented multiple possibilities. Prior studies have emphasized that the implementation of digital transformation has enhanced the effectiveness of organizational procedures and the caliber of public services. Nevertheless, there are several elements that influence the effectiveness of information systems (IS) in the public sector. Users’ satisfaction is considered one of the most important conditions for digital transformation. This article aims to analyze the determinants that impact IS acceptance and users’ satisfaction in the public sector. Data were collected by 498 IS users of many organizations in the Greek public sector and analyzed using Structural Equation Modeling. The results presented in this study offer valuable insights for practitioners involved in the development of such systems. These insights can enhance the effectiveness of their work and encourage them to consider these variables when creating and implementing these systems in the public sector.1612147216
Exploring tactile identification accuracy through advanced braille embossers: The dynamic combination of dot density and dot elevation
No full textThe scope of this paper is to examine the potential capabilities of embossed printers under the lens of Information Technology (IT). One of the main principles of IT is to manage and deliver information into accessible formats to all users with and without disabilities. The present study investigates the discriminability of lines and squares by participants who had severe visual impairment. These lines and squares, produced by braille embossers, represented shapes, figures, and diagrams from a variety of textbooks. Hence, it was expected that the more discriminable these tactile formats are by touch, the more accessible and understandable will be by people who are blind, enhancing the level of collaborative learning and performance. Thirty-four volunteers participated in experiments, and they were invited to conduct matching activities through touch. The researchers, during the experimental phase, recorded (a) all participants’ matching stimuli responses and (b) the time taken for each identification during the matching activities. The tactile stimulus material - tactile lines and square areas - was produced by an advanced braille embosser. Four levels of dot heights (0.030, 0.079, 0.227, and 0.468 mm, respectively) were combined with two versions of densities (10 and 20 dpi, respectively) to create embossed dotted lines and square areas. The results indicated that lines had roughly the same success proportion at the two resolutions (10dpi vs. 20dpi). However, squares had a significantly greater success proportion at 10dpi than at 20dpi. In fact, squares were better than lines at 10 dpi but worse than lines at 20 dpi. Finally, the present study highlights the need to examine further the discriminability among the combinations of embossed lines and areas in order to have a clear-cut “picture” of what is really perceived by participants who are blind and then attempt to develop and propose a standardized protocol to incorporate it in the assistive technology market and in the educational reality as well. Many researchers hold the view that apart from the increased accessibility of students who have visual impairments in the content of their classes, communication and discussions will also be enhanced because of the increased interaction and collaboration among students. According to the researchers, this outcome may enhance the dynamics of an inclusive setup, ensuring a potential well-balanced social integration. Finally, it is argued that the production of outstanding and fine tactile graphs will enable students with vision disability to have access to relevant educational content (such as in science and mathematics), ensuring equity and effective inclusion.2918253332535