To enhance the efficacy of traditional English enlightenment education for children, this research delves into a multi-modal teaching approach and incorporates a hidden Markov model to refine the precision of speech recognition. Within the input recognition module, voice input is combined with operational input, resulting in a multi-modal fusion perception module designed to amalgamate students’ learning and operational intent. Concurrently, a multi-modal natural interaction module for intention understanding is formulated to augment the quality of interaction during the teaching. The research findings revealed that the accuracy of the speech input recognition model surpassed that of the conventional model, increasing from 57.14 to 71.05
This paper briefly outlines current literature on evolutionary architectures and current links with microservices orchestration and data integration. We also propose future research directions bridging the field of service-oriented architectures with the data science domain.
Accurate diagnosis and treatment planning for medical conditions rely heavily on the results of medical image segmentation. Medical images are available in many modalities like CT scans, MRI, histopathological, and ultrasound images. Among all, the real-time analysis of the ultrasound is the most complex as the internal organ’s visualization requires experience from the radiologist. Diagnosing the medical conditions and unavailability of experienced radiologists during an emergency requires automated segmentation which heavily depends on computer-aided diagnostic systems. The new generation CAD systems are found to incorporate advanced deep learning algorithms to produce accurate segmentation results. While most of the segmentation models relate to the encoder-decoder model as the base architecture and thus evolve a variety of modifications in its pipeline architecture. This paper presents the analytical study of the various Encoder- Decoder based models like UNet, Residual UNet (Res-U-Net), Dense UNet (DenseUNet), Attention UNet, UNet + +, Double UNet, and U 2 Net (U-Squared-Net) on ultrasound image segmentation. Further, the paper presents the various trade-offs, application areas, open challenges, and performance analysis of these models on benchmark datasets, namely the HC18 Challenge dataset, CUM dataset, and B-mode Ultrasound Nerve Segmentation dataset. The performance analysis of these models is presented using the six state-of-the-art metrics like Dice coefficient, Jaccard index, sensitivity, specificity, Mean Absolute distance, and Housdorff Distance. Based on the above parameters U 2 -Net (U-Squared-Net) outperformed all other neural network models for all three datasets. In terms of all four criteria (Dice Coefficient: 0.92, 0.89, 0.9, Jaccard Index: 0.81, 0.79, 0.81, Sensitivity: 0.86, 0.84, 0.86, Specificity: 0.97, 0.95, 0.96), the U2-Net (U-Squared-Net) model performed the best. Over the HC18 Challenge dataset, the CUM dataset, and the B-Mode Ultrasound nerve segmentation dataset, U2-Net (U-Squared-Net) model achieved the best HD results (HC18: 3.8, CUM Dataset: 4, B-mode US Dataset: 3.6) and the lowest MAD values (HC18: 2.1, CUM Dataset: 3, B-mode US Dataset: 2.15). In addition, the analysis also highlights the architectural differences between these models, focusing on their type of connections, number of layers, and additional components. The outcome of this research provides valuable perception into the strengths and limitations of each encoder-decoder-based model, aiding researchers and practitioners in selecting the most appropriate model for Ultrasound image segmentation tasks.
Service providers provision more and more Internet-of-Things (IoT) services in the cloud for dynamicity and cost-effectiveness purposes. This is made possible thanks to the introduction of edge computing that brings additional computing and resources for analytics close to the data sources and thus enables meeting the low latency requirement. Edge nodes should support (i) the heterogeneity of IoT devices (e.g., sensor, actuator) and (ii) characteristics (e.g., mobility, location awareness). IoT is already integrated to the hybrid cloud/edge environment. However, the ecosystem lacks of automation due to the previously mentioned characteristics. Indeed, edge nodes are often manually selected during deployment time, and most of the regular quality-of-service (QoS) management procedures remain difficult to implement. This paper introduces a comprehensive semantic model called EdgeOnto. It encompasses all concepts related to IoT applied in the context of edge computing. The ultimate goal of EdgeOnto is to automate the several steps that make up the IoT services lifecycle in hybrid cloud/edge environment. On the one hand, semantics enable an automatic discovery of the relevant edge nodes that are suitable to host and execute IoT services considering their requirements. On the other hand, it allows supporting the specific QoS procedures that are related to such setting (e.g., low latency, mobility, jitter). The core ontology was designed with the Protégé open-source tool. A smart strawberry farming use case was implemented and evaluated for illustration purposes. The results validate the accuracy and the precision of the designed semantic matchmaker.
Public authorities worldwide provide a large number of Public Services (PS) to citizens, businesses and other authorities. In this context, they publish PS catalogues containing descriptions of these services, e.g. about cost, required documents, contact details, etc. Two main challenges in the design of PS catalogues are standardisation and interoperability. To address these challenges, the European Union (EU) has developed the Core Public Service Vocabulary (CPSV), as the proposed EU standard for PS modelling. CPSV-AP is an application profile of CPSV that uses linked data as an underpinning technology to exploit its benefits, e.g. interlinking PS descriptions with each other and with other web of data resources. Despite its potential however, there is only limited research on CPSV-AP use in practice. The aim of this paper is to devise a process for using CPSV-AP and conduct a pilot implementation, using this process, to investigate potential benefits or challenges from CPSV-AP practical usage. In the framework of this pilot implementation, we publish a set of PS descriptions, included in an official PS catalogue, as CPSV-AP compliant linked data. The contribution of this paper is a detailed process for CPSV-AP use in practice and relevant lessons learnt. We anticipate this research will be beneficial to both academics working on PS semantic interoperability and practitioners aiming to migrate existing PS catalogues exploiting CPSV-AP.
Enterprises rely on their process-aware information system (PAIS) to conduct business. Therefore, appropriately responding to environmental changes is vital for enterprises to maintain competitiveness. However, one type of change, namely the long-tailed change (LTC), has been overlooked by traditional business process management practice because of its variety and infrequency. Just as the long-tailed effect reveals, the impact of LTC on enterprise PAIS might be no less dramatic than the impact of high-frequency changes. Since business process models are core assets of an enterprise, it is profitable to reuse them efficiently while tackling the conflict of flexibility and applicability in a timely way. This paper proposes a process model maintenance approach to responding to LTCs. By supporting business analysts to add syntax-correct annotations to existing business process models, the approach achieves an agile, error-free, and low-cost mechanism for dealing with LTCs.
Cloud computing has recently evolved into container technology and edge computing for conveniently hosting and delivering a set of services. There is a need to compose individual services to perform a complex task. However, finding an optimal solution of composition satisfying both functional and non-functional requirements is a challenging problem. To solve this problem, we propose a method using fuzzy analytical hierarchy process and an extended Dijkstra algorithm to find a solution in a directed bipartite graph stored in a Neo4j graph database. Unlike other database-based approaches that aim to optimize a pre-defined composition task, our approach is much flexible that can perform an arbitrary task once the services information is preprocessed in a database. We conduct experiments and compare the performance of the proposed approach with a Graphplan method and a relational database-based approach, and the results show that this approach leads to less query execution time and better user's satisfaction.
Web APIs have been adopted as the de facto standard for exchanging data on the Web. However, engineering applications that orchestrate the invocation of multiple APIs and the data flow among them are still mostly manual and labor intensive. In fact, as the number of the potentially relevant APIs increases, compositions become opaque, difficult to maintain, and practically impossible to reuse. The recent advances around linked data formalisms have the potential to provide “usable” semantics, to enable automatic API composition methods. In this paper, we formalize a simplified description model, based on SPARQL graph patterns, for capturing the semantics of Web APIs. Based on this model, we propose a methodology for a fully automated process that produces semantically valid composition chains, using iterative subgraph isomorphism. We have validated the usefulness and accuracy of our approach, using a collection of publicly available Web APIs relevant to a real-world use cases.
Virtual world has the potential to become a future global electronic marketplace, integrating many isolated markets in many areas. To achieve this goal, future virtual world is required to be persistent, implying that a virtual world together with its accumulated content shall exist forever regardless of dynamic changes of its users and owners. Unfortunately, existing virtual worlds, owning by some entities, are not immune from death due to business entity failure. To provide a persistent virtual world, a decentralized architecture is explored, which is constructed on user contributed devices. However, there are many challenges to realize a decentralized virtual world. One important issue is user cooperation in reliable content storage. The devices contributed by users may not be reliable for maintaining all user contents, but users do not have the incentive to provide reliable devices for others. This paper addresses the issue by two steps. First, an indicator, called replica group reliability, is provided to users, which is based on the proposed replicability index. Based on the indicator, users can learn the reliability of their content storage. Then, a new user incentive mechanism, called equity-based node allocation strategy, is proposed to promote user cooperation to collectively maintain reliable content storage. A decentralized algorithm implementing the strategy is designed and the evaluation results show its effectiveness and efficiency.
Selecting an enterprise architecture framework (EAF) that will best address the needs of their organization is a difficult task for enterprise architecture practitioners. The objective of this study is to make this difficult task easier. To do so, this research first conducts a review of the literature on EAF evaluation criteria. Findings from this review show the shortcomings of this literature, most importantly the lack of a comprehensive set of EAF criteria and adequate measures for their operationalization. Based on these findings, and using the design science research approach, this study then designs and tests an EAF evaluation artifact that identifies, elaborates and operationalizes a comprehensive set of 14 criteria. Results of the experiment that followed show that: (i) 90.87% of the criteria were perceived usable, (ii) 97.62% of them were perceived to be applicable and relevant, and (iii) 90.48% were perceived as correct. This study makes several contributions. First, it provides a much-needed and timely overview of the literature on EAF evaluation criteria. Second, this study is the first to present a comprehensive set of criteria that not only synthesizes previously proposed criteria but also includes new criteria (e.g., SOA models and usability). Third, it answers EA practitioners' requests for a tool that is both theoretically sound and practical. Finally, and most importantly, this study is the first to propose objective measures to operationalize EAF evaluation criteria.
Community-built telecommunication networks such as Guifi.net demonstrate how end users can actively collaborate in the self-provision of network services, for instance by operating a self-organized distributed monitoring system. Network monitoring is performed by many small servers at the users' premises but data are only accessible via a centralized interface. Besides, due to network partitions and churn of the monitoring servers, failures in the monitoring system are frequent, leaving parts of the network unmonitored. Distributed databases are a promising solution for data replication under network partition condition, but they suffer from a trade-off between data consistency and availability. Furthermore, these databases are used in data centers with abundant computing resources, not in light edge networks. In this work we present DIMON, a reliable edge-based, eventually-consistent monitoring system that leverages CRDT-based data structures implemented in AntidoteDB. Conflict-free replicated data types (CRDTs) are able to converge to a consistent state in environments with network partitions as those found in edge networks. Our results give insights on the load of AntidoteDB on edge devices under different scenarios of read and write operations. The experiments carried out in a production network with a real system implemented contribute to the research community's knowledge about the available technologies for a consistent replicated data storage layer to support edge computing clouds.
For centuries, crime has been viewed as random because it is based on human behavior; even now, it incorporates an excessive number of factors for current machine learning models to forecast accurately. In this work, we tend to discuss the early crime prediction results from a model developed using the data from the Chicago crime dataset. In any case, with a superior execution future crime is to anticipated accurately, it is a testing assignment as a result of the increase in several crimes in present days. Therefore, the crime foreseeing method is foremost, and it identifies the future crimes and number of crimes are degraded. In this paper, we built up a model to anticipate future crime occurrences at a future time and also predict which type of crime may be happening in a given area. First, we analyze how certain crime features like given a date, time and some geologically important relevant features such as latitude and longitude. Second, we discuss several analytics techniques we used to find meaning in our data, such as LASSO feature selection analysis, classification models like naïve Bayes and SVM. Finally, we select the best model for foreseeing crime type and seriousness of the crime for giving different features.
This study uses the edge computing system for fast quality screening in the metal processing industry, and the Internet of Things technology is responsible for delivering data to the cloud for visualization and analysis. First, the system consists of optical components and embedded systems. Further, a fast Fourier transform is used to make the image have frequency characteristics. However, the convolution operation between the random kernel model and the image is the main means of feature extraction. In order to evaluate the performance and convergence of the proposed method, a rapid screening mechanism for good/bad products is defined. Finally, the data is passed to the cloud (ThingSpeak) platform for visualization through the MQTT protocol, and the content is subscribed to the content by the background host to perform the quality decision of the fuzzy inference system. The result is released back to the cloud. The experimental results in the industrial example show that the proposed method can accurately and quickly complete the quality inspection of surface roughness, and the feature distribution is easy to understand. At the same time, the edge computing system has the advantages of instant response and low cost, while the Internet of Things technology brings more management and analysis convenience. In future research, the unsupervised learning algorithm based on convolutional neural networks is a potential application, which can learn the quality of good or bad through a large amount of data.
Currently, the most commonly used personal authentication system for digital deveces use text-based passwords. However, long passwords which increse the security are hard to remember and decrease the usability. Nowadays, alternative systems are used such as graphical password systems and biometrics systems. Each of them has its own problem also. Lost password cannot be replaced in case of biometric system and graphical password systems are weak at shoulder-surfing attacks. This paper describes implementation of a new personal authentication scheme called T-TIME. T-TIME uses touch signal generation time information which is cumbersome to identify by watching.
As cybersecurity is a growing field of science, there is not complete agreement across the scope of cybersecurity research topics. This paper proposes a grouping and classifications of cybersecurity researches to introduce an easily referenceable taxonomy of the cybersecurity research topics. A literature survey was conducted to collect published literature review papers about various cybersecurity researches during the past five years periods. Taxonomic analysis of 99 selected papers, which were grouped based on research topic similarities, produced sets of research categories. It categorized the cybersecurity research topics into 8 areas: (1) Applied cybersecurity, (2) Cybersecurity data science, (3) Cybersecurity education and training, (4) Cybersecurity incidents, (5) Cybersecurity management and policy, (6) Cybersecurity technology, (7) Human and social cybersecurity and (8) Theories in cybersecurity. Although cybersecurity grew out of the computer science field, this paper argues that its interdisciplinary nature (not only technical computer security, but also data, system/technology, and human/social) can attract researcher from various disciplines, such as management, policy, psychology, and so on, to contribute to cybersecurity advancement.
One of alternative ways to achieve a competitive advantage by an enterprise in the ICT area is to select the most prospective information system (IS) as a seed technology. The latter is defined as a first technology from a given area implemented in an enterprise, that determines or strongly influences the choice of further technologies from the same area. Seed ICT selection is particularly relevant for enterprises whose core business activity relies on online service provision. In this paper we present a user-requirement-based interaction scheme of an application that models the evolution of enterprise ISs, taking into account the production patterns of software releases, the functionalities offered, the users' community growth trends, and other factors. The evolution is modeled by hybrid discrete-time/discrete-event systems, and multi-models built with them. The technology choice is supported by multicriteria ranking algorithms that apply IS innovativeness forecasts. Moreover, this evolution model of enterprise IS can handle different uncertainties and cyber security issues. We present the use cases and scenarios, model management, and the knowledge base that stores time series, IS characteristics and forecasts. The use of this application is illustrated by a real-life example of three popular open-source CMSs. We conclude that an informed choice of the first IS and the corresponding enterprise architecture as a seed technology is particularly relevant for start-ups and SMEs. An extended variant of the above application may also provide decision support to developer teams seeking software evolution models to discover and apply the best-possible technology development and market strategies.
Many IoT systems are data intensive and are for the purpose of monitoring of critical systems. In these monitoring systems, a large volume of data steadily flow out of a large number of sensors which monitor the physical systems and environments. Thus, first of all, we need to consider how to store and manage these IoT data. Also, data sharing can greatly enhance the quality of data analytics and help with cold start of similar systems. Thus, the data storage and management solutions should consider how to help discover useful data in order to facilitate data sharing. Time series databases (TSDBs) have been developed in recent years for storing IoT data, but they have some deficiencies. One problem is that they are not very effective in supporting data sharing due to the lack of a good semantic model for proper data specifications, which is critical in data discovery. To resolve this problem, we develop a monitoring data annotation (MDA) model to guide the systematic specification of monitoring data streams. To support the realization of the MDA model, we also develop an external tool suite, which stores the additional MDA-based specifications for the data streams and interfaces with queries to perform preliminary processing to allow effective monitoring data discovery based on the MDA specifications. Another problem with current TSDBs is their focus on storing time series data that arrive at a fixed rate, but not on storing and retrieval of event data, which may come sporadically with irregular timing patterns. When storing such event data in existing TSDBs, the retrieval may have performance problems. Also, existing TSDBs do not have specific query language defined for event analysis. We develop a model for event specifications and use it to specify abnormal system states to be captured to allow timely mitigation. The event model is integrated into the TSDB by translating them to continuous queries defined in some TSDBs. Also, we develop an event storage scheme and incorporate it in TSDBs to facilitate efficient event retrieval. Experimental results show that our event solution for the TSDB is effective and efficient.
The emergence of IoT and AI has brought revolutionary change in various application domains. One of them is Industry 4.0, also called Smart Manufacturing, which aims to achieve highly flexible and automated production processes. In this paper, we study a use case of anomaly detection in smart manufacturing using the real data collected from the sensing devices of a factory production line. Our goal is to improve the anomaly detection accuracy at an earlier stage of production line, so that cost and time wasted by possible production failures can be reduced. To overcome the limited and irregular anomaly patterns found from our multivariate sensor dataset, we proposed an unsupervised real-time anomaly detection algorithm based on LSTM-based Auto-Encoder. Our evaluations show that our approach achieved almost 90% accuracy for both precision and recall while other classification or regression based methods only reached 70%~85%.
This paper proposes a sport-service-oriented architecture framework and describes an example of design and implementation of an IoT application in sports based on sport-skills researches. Especially, the focus is on wearable applications with inertial sensors because sports applications require simple and durable motion-sensing systems. Skateboarding is introduced as a target sport to understand skill components and to support skill acquisition and improvement. Two types of skills (i.e., basic and advanced) are introduced to show different requirements for IoT applications in sports. Design factors to implement a sports application are discussed in terms of two aspects (i.e., analysis and service phases) concerning sports activities. To implement these designs, technology selection and implementation alternatives are discussed in terms of skill types and purpose of sensing (i.e., analysis or service phase).
