According to the advances in users' service requirements, physical hardware accessibility, and speed of resource delivery, Cloud Computing (CC) is an essential technology to be used in many fields.Moreover, the Internet of Things (IoT) is employed for more communication flexibility and richness that are required to obtain fruitful services.A multi-agent system might be a proper solution to control the load balancing of interaction and communication among agents.This paper proposes a multi-agent load balancing framework that consists of two phases to optimize the workload among different servers with large-scale CC power with various utilities and a significant number of IoT devices with low resources.Different agents are integrated based on relevant features of behavioral interaction using classification techniques to balance the workload.A load balancing algorithm is developed to serve users' requests to improve the solution of workload problems with an efficient distribution.The activity task from IoT devices has been classified by feature selection methods in the preparatory phase to optimize the scalability of CC.Then, the server's availability is checked and the classified task is assigned to its suitable server in the main phase to enhance the cloud environment performance.Multi-agent load balancing framework is succeeded to cope with the importance of using large-scale requirements of CC and (low resources and large number) of IoT.

—Traditional machine learning (ML) techniques model knowledge using static datasets. With the increased use of the Internet in today's digital world, a massive amount of data is generated at an accelerated rate that must be handled. This data must be handled as soon as it arrives because it is continuous, and cannot be kept for a long period of time. Various methods exist for mining data from streams. When developing methods like these, the machine learning community put accuracy and execution time first. Numerous sorts of studies take energy consumption into consideration while evaluating data mining methods. However, this work concentrates on Very Fast Decision Tree, which is the most often used technique in data flow classification, despite the fact that it wastes a huge amount of energy on trivial calculations. The research presents a proposed mechanism for upgrading the algorithm's energy usage and restricts computational resources, without compromising the algorithm's efficiency. The mechanism has two stages: the first is to eliminate a set of bad features that increase computational complexity and waste energy, and the second is to group the good features into a candidate group that will be used instead of using all of the attributes in the next iteration. Experiments were conducted on real-world benchmark and synthetic datasets to compare the proposed method to state-of-the-art algorithms in previous works. The proposed algorithm works considerably better and faster with less energy while maintaining accuracy.

According to the advances in users' service requirements, physical hardware accessibility, and speed of resource delivery, Cloud Computing (CC) is an essential technology to be used in many fields. Moreover, the Internet of Things (IoT) is employed for more communication flexibility and richness that are required to obtain fruitful services. A multi-agent system might be a proper solution to control the load balancing of interaction and communication among agents. This paper proposes a multi-agent load balancing framework that consists of two phases to optimize the workload among different servers with large-scale CC power with various utilities and a significant number of IoT devices with low resources. Different agents are integrated based on relevant features of behavioral interaction using classification techniques to balance the workload. A load balancing algorithm is developed to serve users' requests to improve the solution of workload problems with an efficient distribution. The activity task from IoT devices has been classified by feature selection methods in the preparatory phase to optimize the scalability of CC. Then, the server's availability is checked and the classified task is assigned to its suitable server in the main phase to enhance the cloud environment performance. Multi-agent load balancing framework is succeeded to cope with the importance of using large-scale requirements of CC and (low resources and large number) of IoT.

Traditional machine learning (ML) algorithms use static datasets to model knowledge. Nowadays, there is an increasing demand for machine learning based solutions that can handle very huge amounts of data in the shape of streams that never stop. The Very Fast Decision Tree (VFDT) is one of the most widely utilized data stream mining algorithms (DSM), despite the fact that it wastes a huge amount of energy on trivial calculations. The machine learning community has come first in terms of accuracy and execution time while designing algorithms like this. When assessing data mining algorithms, numerous types of studies include energy usage as a crucial factor. The purpose of this research is to create a hyper model to optimize the VFDT algorithm, which reduces the waste of energy while maintaining accuracy. In the proposed method, some fixed algorithm parameters were changed to dynamic parameters after analyzing each of them separately and knowing the extent of their positive impact on reducing energy consumption in several cases in algorithm. The practical experiment was conducted on both the algorithm in its basic form and the algorithm in the proposed form on several different types of datasets in the same application environment The main advantage of the results of the proposed method compared to the results of the basic algorithm is that there was a noticeable development in the performance of the algorithm in terms of reducing its energy consumption and maintaining its accuracy levels.

Recently, artificial intelligence (AI) domain increased to contain finance, education, health, mining, and education. Artificial intelligence controls the performance of systems that use new technologies, especially in the education environment. The multiagent system (MAS) is considered an intelligent system to facilitate the e-learning process in the educational environment. MAS is used to make interaction easily among agents, which supports the use of feature selection. The feature selection methods are used to select the important and relevant features from the database that could help machine learning algorithms produce high performance. This paper aims to propose an effective and suitable system for multiagent-based machine learning algorithms and feature selection methods to enhance the e-learning process in the educational environment which predicts pass or fail results. The univariate and Extra Trees feature selection methods are used to select the essential attributes from the database. Five machine learning algorithms named Decision Tree (DT), Logistic Regression (LR), Random Forest (RF), Naive Bayes (NB), and K-nearest neighbors algorithm (KNN) are applied to all features and selected features. The results showed that the learning algorithm that has been measured by the Extra Trees method has achieved the highest performance depending on the evaluation of cross-validation and testing.

Deep learning has become one of the crucial trends in the modern era due to the huge amount of data that has become available. This paper aims to investigate and improve a generic framework for Arabic Text Classification (ATC) with different deep learning techniques. Besides, it deals directly with a word in its original style as a basic unit of modern Arabic sentence and on a different level of N-grams versus a combination of Intersected Consecutive Word proposed method (ICW). However, it aimed to discuss the results of the different experiments for the enhancements of the proposed method on different deep learning algorithms such as Scaled Conjugate Gradient (SCG) and Gradient descent with momentum and adaptive learning rate backpropagation (GDX) on ATC. The results showed that the proposed framework applied with the SCG algorithm and TF-IDF outperforms the GDX algorithm with an accuracy ratio of 90.65%.

In data mining, outlier detection is applied in different domains. It has very large applications such as energy consumption analysis, forecasting hurricanes in meteorological data, fraud and intrusion detection, event detection and system monitoring in sensor networks, etc. Most of existing outlier detection techniques depend on the properties of a particular type of data and can not deal with a large volume of data well, which mean that there is a necessity for improved methodologies and techniques to be applied to a large amount of data with different types in other application areas. In this paper, a parallel outlier detection technique is developed to detect the outliers in the sequential data. Although there are many types of outliers, this paper concentrates on the contextual anomalies. The proposed technique uses a graph approach to detect the outliers. It is very flexible, fast, and no labeled data is needed comparing to many previous approaches. The experimental results show the detected contextual outliers in the sequential data, as well as the efficient scaling up to handle the massive data by increasing the number of processors. The results prove that the parallelism of the proposed technique is very valuable.

The Coronavirus disease (COVID-19) pandemic has led to a huge loss of human life. It has also severely affected the economic, social, and health systems around the world. Frequent pattern mining is one of the main research topics in data stream mining. It is significant in many critical applications, especially in the medical field. This paper proposes a Compressed Maximal Frequent Pattern based on a Damped Window model over a data stream (CMFP-DW). Its main contribution is to integrate the concept of correlation with the purpose of finding valuable patterns that are highly correlated. As such, a new type of pattern is defined, namely the correlated compressed maximal frequent pattern. The CMFP-DW approach is employed for mining accurate correlated maximal frequent patterns from streaming data, and it has been validated against a real-world COVID-19 dataset from the healthcare domain. Frequent patterns generated from this dataset are exploited with the purpose of detecting the COVID-19 cases in different countries of the world. This helps decision makers take the appropriate precautions to prevent the further spread of the COVID-19 pandemic across the world. The six experiments carried out show that the proposed approach outperforms two other existing approaches, namely the estDec and the CP-Tree algorithms regarding accuracy in extracting correlated maximal frequent patterns, memory usage, and the required response time.

In recent years, there has been increasing interest in cloud computing research, especially replication strategies and their applications.When the number of replicas is increased and placed in different places, maintaining the system's data availability, performance and reliability will increase the cost.In this paper, two multi-objectives swarm intelligence algorithms are used to optimize the data replication selection and placement in a cloud environment.These algorithms are namely, multi-objective particle swarm optimization (MOPSO) and multi-objective ant colony optimization (MOACO).The first algorithm, (MOPSO), is used to find the best selected data replica according to the most popular data replication strategy.The improved time-based decay function (ITBDF), is used to enhance the proposed model.The second algorithm, (MOACO), is used to find the best data replica placement according to the minimum distance, the number of data transmissions and the availability of data replication.A simulation of the suggested strategy has been performed using CloudSim.the Cloud is formed to simulate different kinds of datacenters (DCs) with different structures.Moreover, 21 DCs are used.Each DC consists of a host that contains a set of virtual machines (VMs) that provides blocks of available data replications.Three different data placements for high datacenters were created.A total of one thousand cloudlets are randomly confirmed for the data replication order.All replication files are placed in high datacenters and randomly distributed in the suggested system.The performance of proposed strategy was evaluated relative to many well-known strategies such as, Enhance Fast Spread (EFS), Dynamic Cost-aware Re-replication and Re-balancing Strategy (DCR2S), Genetic Algorithm (GA), Genetic adaptive Selection Algorithm (GASA), Replica Selection and Placement (RSP), Dynamic Replica Selection Ant Colony Optimization (DRSACO), Adaptive Replica Dynamic Strategy (ARDS), Popular File Replication First (PFRF).The experimental results show that MOPSO, achieves better data replication than compared algorithms.Additionally, MOACO, achieves higher data availability, lower cost, and less bandwidth consumption than compared algorithms.

Research in frequent pattern mining from streaming data becomes a pioneer in the field of information systems. The data stream is a continuous flow of data generated from different sources. Extracting frequent patterns from streaming data raises new challenges for the data mining community. We present an overview of the growing field of data streams. Many applications handle streaming data such as sensor networks, traffic management, log data, telephone call records, and social networks. These applications generate high volumes of streaming data with velocity, which is difficult to handle with traditional data mining techniques. This paper mainly reviewed different research algorithms, scientific practices, and methods that have been developed for mining frequent patterns from streaming data. In addition, it discusses well-known open-source software and tools for data stream mining, which are developing to handle streaming data. Finally, it summarizes the open issues and challenges to current existing approaches while handling and processing data streams in realworld applications. Keywords— Data streams, Frequent pattern mining, Stream data mining, Concept Drift, Window models.

Large graphs are available in everywhere such as social networks, web link analysis, and computer networks.Traditional methods of clustering are not suitable to tackle the problem of clustering large graphs because the computation is very costly, which is solved by local graph clustering using a given vertex set as input to detect an accurate cluster.SimpleLocal (SL) algorithm detects a best conductance cuts close to seed vertices set.In this paper, a new Parallel SimpleLocal (PSL) system is proposed using multicore CPUs.OpenMP parallel library is utilized to parallelize the first and second stages of 3StageFlow algorithm whereas the SL algorithm is used for enhancing the runtime.The experiments are performed on various applications from different domains, which are image segmentation and community detection.From the experiments, the proposed method improves the runtimes with 75.43% using 4-cores and 81.01%when using 8-cores over the sequential single core.

In recent years, cloud computing research, specifically data replication techniques and their applications, has been growing. If the replicas number is raised and put in multiple positions, it will be expensive to maintain the data usability, performance and stability of the application systems. In this paper, two bio- inspired algorithms were proposed to improve both selection and placement of data replicas in the cloud environment. The suggested algorithms for dynamic data replication are multi-objective particle swarm optimization (MO-PSO) and ant colony optimization (MO-ACO). The first suggested algorithm, i.e ., MO-PSO, is employed to obtain the best selected data replica depend on the most frequent one. However, the second suggested algorithm, i.e ., MO-ACO, is employed to obtain the best data replica placement depend on the shortest distance, and the replicas availability. A simulation of the suggested strategy was carried out using CloudSim. Each data center (DC) includes hosts with set of virtual machines (VMs). The data replication order is determined at random from a thousand cloudlets. All replication files are randomly distributed in the proposed architecture. The performance of suggested techniques was evaluated against several approaches including: Adaptive Replica Dynamic Strategy (ARDS), Enhance Fast Spread (EFS), Genetic Algorithm (GA), Replica Selection and Placement (RSP), Popular File Replication First (PFRF), and Dynamic Cost-aware Re-replication and Re-balancing Strategy (DCR2S). The simulation results prove that MOPSO gives improved data replication compared against other algorithms. Additionally, MOACO realizes higher data availability, lower cost, and less bandwidth consumption compared with other algorithms.

In the educational field, the system performance, as well as the stakeholders’ satisfaction, are considered a bottleneck in the e-learning system due to the high number of users who are represented in the educational system’s stakeholders including instructors and students. On the other hand, successful resource utilization in cloud systems is one of the key factors for increasing system performance which is strongly related to the ability for the optimal load distribution. In this study, a novel load-balancing algorithm is proposed. The proposed algorithm aims to optimize the educational system’s performance and, consequently, the users’ satisfaction in the educational field represented by the students. The proposed enhancement in the e-learning system has been evaluated by two methods, first, a simulation experiment for confirming the applicability of the proposed algorithm. Then a real-case experiment has been applied to the e-learning system at Helwan University. The results revealed the advantages of the proposed algorithm over other well-known load balancing algorithms. A questionnaire was also developed to measure the users’ satisfaction with the system’s performance. A total of 3,670 thousand out of 5,000 students have responded, and the results have revealed a satisfaction percentage of 95.4% in the e-learning field represented by the students.

Cloud computing is a modern technology for dealing with large-scale data. The Cloud has been used to process the selection and placement of replications on a large scale. Most previous studies concerning replication used mathematical models, and few studies focused on artificial intelligence (AI). The Artificial Bee Colony (ABC) is a member of the family of swarm intelligence based algorithms. It simulates bee direction to the final route and has been proven to be effective for optimization. In this paper, we present the different costs and shortest route sides in the Cloud with regard to replication and its placement between data centers (DCs) through Multi-Objective Optimization (MOO) and evaluate the cost distance by using the knapsack problem. ABC has been used to solve shortest route and lower cost problems to identify the best selection for replication placement, according to the distance or shortest routes and lower costs that the knapsack approach has used to solve these problems. Multi-objective optimization with the artificial bee colony (MOABC) algorithm can be used to achieve highest efficiency and lowest costs in the proposed system. MOABC can find an optimal solution for the best placement of data replicas according to the minimum distance and the number of data transmissions, affording low cost with the knapsack approach and availability of data replication.Low cost and fast access are characteristics that guide the shortest route in the CloudSim implementation as well. The experimental results show that the proposed MOABC is more efficient and effective for the best placement of replications than compared algorithms.

Nowadays, large number of applications of graph clustering are available, with expanding the span of the graph the conventional methods of clustering are not appropriate to manipulate this issue which are costly for computation. So that, it is necessary to get a good algorithm to tackle this problem. Graph clustering algorithms are considered as the most effective techniques for solving various partitioning problems. Global graph clustering which based on the whole graph as input isn’t convenient of large graphs. Local graph clustering algorithms solve this problem by working on a given vertex as input seed set without looking at the whole graph to find a good cluster. This research explores different graph clustering techniques based on the input parameters, e.g., local and global, as well as illustrating appropriate applications of graph clustering. This paper directed to help new researchers take a summary of graph clustering techniques that can be used for graph partitioning.

Data are the representation of our world and our life. Data are increasing continuously, they come from different sources such as sensors, maps, climate informatics, smartphones, social media and/or medical data domains. Data are represented by different forms such as image, text, video and/or digital data. These incomprehensible data need an influential technique to be clustered and analyzed. This paper presents a hashing technique for the clustering process of unclassified and disorganized data. These clustered data are useful for decision-making process. The proposed technique is based on Golay error-correction code. The main concept is reversing the original Golay error-correction scheme and building Golay Code Addresses Hash Table (GCAHT). Simulation results stated that the proposed technique achieved high performance. Beta-CV, Dunn Index, C-index and Sum Square Error are used for measurements.

Categorisation of text is significant trend that ultimately appears owing to the internet revolution nowadays resulting in enormous amounts data that depend on various languages. The Arabic language is one of the most commonly used languages all over the world; it is considered the fifth most spoken one. Various challenges occur through processing and classifying of Arabic text since it has more sophisticated techniques than the English language. These challenges are clear owing to the Arabic language's variation in shape, structure and component; besides, there is a lack of adequate studies discussing Arabic text classification. This research seeks to form a general point of view by categorising different techniques of Arabic text classification for helping new researches concerning this domain. Also, it shows some of prior information and innovative designs about Arabic text classification. Besides, it mentions various works that have discussed classifying Arabic text, with regard to data sets, categories and pre-processes steps, classification mechanism and assessment procedure for those techniques. These discussions aim to conclude a comprehensive overview through forming a general framework for all researchers about this domain via examining the defects of the prior studies, and then the possibility of presenting more advanced directions.

Software release planning is one of the complicated areas of software engineering. Planning a software release includes assigning requirements to sequence of releases in the most beneficial way within the limited effort, budget, and time available. The complexity of the software release planning is due to the incompleteness and the uncertainty challenges that characterise the software release planning problem. In order to handle efficiently the incompleteness and uncertainty challenges, the computer intelligent fuzzy logic approach can be utilised. Although several methods have been proposed to handle the incompleteness and uncertainty challenges, they are all based on optimisation methods. The objective of this study is to propose a novel framework to handle the uncertainty and incomplete challenges of the software release planning based on intelligent approaches such as fuzzy rule-based system rather than optimisation method. Also, some predefined thresholds are determined to support the time flexible horizon challenge of the software release planning process and to increase the flexibility of the proposed framework. For validating the proposed framework, it is applied to plan a new release for updating a faculty website project.

In data area, to achieve good information for decision making, suitable processing of data is needed. Data need to be transferred. They are transferred as a vector which contains features of data. During data transferring, errors may occur. Errors change the features of data vector (instance). In this case, error detection and correction techniques are needed to tackle this issue. If data transferred as groups based on its features, any change in the features of any vector will change the group (cluster) of this vector. So, to cluster an incomprehensible data for operating any method of data mining, an influential technique is needed, and this technique should ensure the correctness of the cluster using error detection and correction codes like Hamming and Golay. This paper presents a technique to detect and correct clustered data after the transfer process to reduce the misclustered instances. The main concept is the reemploying the error-correction Golay code with splitting of the data word and code word to symbols. DENCLUE clustering algorithm is used in the step of clustering as density-based clustering algorithm. Comparison with other related works is performed and the simulation results stated that the proposed technique achieved better performance.

Recently, number of applications including social networks, stock market trading and sensor network devices generate a massive amount of data in the streaming form. Streaming data have characteristics different from static data, such as streaming data arrives continuously at high speed with huge amount. Mining and discovering information from these data is a non-trivial issue. Most of traditional algorithms have limitations to deal with streaming data, so there are new issues raised and need to be taken into consideration while developing techniques for mining association rules from such data. In this paper, a technique to mine an association rules from streaming data efficiently is proposed. The proposed technique develops a tree structure called Fast Update Frequent Pattern Tree (FUFP-Tree) that reduce the number of traversing between tree nodes in both inserting a new transaction and extracting an association rules between items. Also, to avoid congestion during inserting incoming streaming data to FUFP-Tree, a sliding window approach is used to divide incoming data equally to all available windows. The complexity and the performance of this technique are investigated, and a dataset of storehouse is used to test the proposed technique and measure its efficiency. The efficiency of the proposed technique is compared with other most related algorithms.