The automated essay scoring (AES) has significant importance in machine grading of student essays particularly in standardized exams like the Graduate Record Examination (GRE). However, some issues in AES have remained unsolved over the past several years. The current approaches have scrutinized AES from both classification and regression perspectives. This study discusses the cutting edge architectures such as RoBERTa, XLNet, and BERT and compares their automated essay scoring performance. ASAP, a publicly accessible dataset is used for this purpose. The obtained results indicate that the natural language understanding (NLU) model proposed in this paper depicts significantly improved performance than all the other existing approaches.
Understanding learner behavior is the key to the success of any learning process. The more we know about learners, the more likely we are to personalize learning experiences and provide successful feedback. This paper presents a feedback rules model called SECA: (i) Scenario, that defines the context behavior in a microlearning environment, (ii) Event, provided by a predictive model, (iii) Condition, that evaluates the events, and (iv) Action, that provides the learner’s feedback. The proposal is achieved through a controlled experiment in which a microlearning environment is available to collect data from a ubiquitous context, and predictive analytics are applied to guide the definition of a set of feedback rules intended to support the learner’s learning process. In the end, we presented an exemplified set of feedback rules, which could be used to provide automatic recommendations and improve the learner experience. Thus, the experiment allows us to analyze the learner behavior in a ubiquitous microlearning context from a feedback perspective.
This study provides a review of solar energy in Iraq, as Iraq is one of the oil-rich countries that are considered more intelligent in the field of alternative energy for the post-oil era, especially as it is located near the solar belt, which makes the solar radiation of high intensity and brightness a period of the year. We must replace non-renewable energy resources (traditional or fossil fuels) with renewable (sustainable) energy resources. This renewable energy (solar energy) is possible, clean, unlimited and environmentally friendly, and it can be used in many applications of lighting, water heating and heating in the winter, and this Reduces the electricity needed during winter for this application. And if we talk about Mosul, the climate of Mosul It’s marked by high summer and cold temperatures .degrees Celsius in the middle of the day, and in December, January and February, temperatures range from -1 degrees Celsius to 8 degrees Celsius. As Iraq is suffering from electricity shortages during this time, so the Iraqi government and the local government of Mosul in particular must take serious decisions and steps to confront and overcome these challenges, and develop developed strategies, and programs of specialized and expert people to meet the increases in demands on electric power. Renewable energies such as solar and wind energy which could play a significant role in Iraq’s future , especially the solar energy covered in this study.
The agricultural sector is unpredictable and complicated for farmers to manage it, especially with all the immoderate challenges that condemn the development of agricultural production. Regarding Morocco, water scarcity remains the major challenge that face the sector and put at risk the irrigation based fields-that represents approximately the half of the agricultural GDP-, the nutritional needs of the Moroccan population as well as the global sustainable growth. A way to handle this issue is adopting green IT that enable farmers to establish one of the latest trends in the global smart irrigation systems which are Green Internet Of Things or G-IoT driven smart irrigation. These technologies are currently highly requested and used in the Moroccan agricultural context contributing to the economic and environmental performance. Consequently, the aim of this study is to inspect the factors that affect the farmer's intention to adopt Green IoT based irrigation systems in the Moroccan case. The study is conducted on a literature review based on studies concerning the same subject matter in an international level, and contribute to the research development by proposing a theoretical model related to farmers’ Green IoT adoption.
Detecting communities of common behaviors, interests, and interactions in social networks is essential to model a network's structure. Overlapping community detection is an NP-Hard problem. Several solutions have been proposed; however, most of these techniques are computationally expensive. We have developed a fast-hierarchical algorithm using the notion of segmentation by weighted aggregation. Experimental results on synthetic and real benchmark networks show that the proposed algorithm effectively finds communities (Clusters) with varied overlap and non-exhaustiveness structures. Our method outperforms the state-of-the-art hierarchical clustering algorithms measured by the F-measure and the computational time.
The Corona pandemic has become a threat to humanity, as the spread of the coronavirus has led to the closure of many institutions as well as many economic, social, and educational activities. In order not to stop the educational process, various educational institutions, including universities, have moved to remote electronic education methods (e-learning) instead of the well-known traditional education system. As several organizations concerned with education have made their electronic platforms and libraries available almost free of charge to alleviate the load on, students and teachers. Despite the advantages provided by e-learning methods, it is accompanied by some difficulties and obstacles that must be dealt with to reduce its effects. This study aims to identify the most important challenges facing the use of e-learning from the university instructor's perspective. The study assesses instructor's opinions, level of satisfaction, and readiness to use e-learning and its platforms available in Libya. A descriptive method was conducted using a questionnaire that was sent electronically to many university instructors to determine the most important challenges they face during the closure as a result of the spread of the Corona pandemic. The study population consists of instructors belonging to six faculties of the Libyan International Medical University. Statistical analysis of the data was performed using SPSS software. The results show that the aspect with the greatest impact was the challenge of the instructors, with the materiality of 69.9% and a mean of 3.49. Social issues ranked second in terms of their impact on e-learning, and all issues related to students came in third place. After that the aspect related to accessibility, which had the least impact on the e-learning process, with the materiality of 55.9 % and a mean of 2.80.
The Internet of things (IoT) is a new technology that shapes the future of a world that is rapidly being invaded by smart devices connected to the Internet. Such technology has a great role in developing the idea of a smart city. A smart city is a city that takes advantage of existing infrastructure and integrates it with the Internet of things technology to improve the quality of life. Internet of Things (IoT) sensors are distributed geographically around the city to collect data from the environment (i.e.: streets, cars, traffic lights...etc.), process, and manage it to provide intelligent actionable information to citizens. All data transferred through networks of a smart city may be threatened and susceptible to illegal actions such as violation, stealing, and inappropriate use. These security threats affect the privacy and security of users; where hackers can get access to user's data and gain control of their smart homes, cars, medical devices and might even gain control over city traffic lights. All the above enforce the need to have a security system that continuously monitors and tracks all data logs to detect any suspicious activity. In this paper, we propose a Security Information and Event Management (SIEM) approach for smart cities by forwarding event logs generated by smart devices to a security operation center that works around the clock to detect security incidents and handle them. Such an approach aims to create a safe smart living environment.
Identifying intrusion in networks is one of the important concerns in computer networks. The task of dimensionality reduction and choice of classifier plays an important role in network intrusion detection. Dimensionality reduction should make sure that the efficacy of classifier on reduced dimensionality data is atleast retained if not improved. In this paper, we suggest a similarity function which can be used to find similarity between any two network elements expressed as vectors. The similarity measure is designed to make sure that the attribute distribution is taken into account for finding similarity value.
Machine learning and statistics are categorized as part of data science. Regression is one of the techniques of machine learning. Most of the contributions in the literature in respect to intrusion detection are mainly based on dimensionality reduction using techniques such as PCA, SVD, feature selection, feature reduction techniques and application of classifier algorithms. Very less attention is paid on regression analysis for intrusion detection in the existing literature. There is a scope to apply regression-based analysis for intrusion detection. Regression analysis may be applied for dimensionality reduction, classification or prediction tasks. This paper throws light on the possibility of applying regression analysis for intrusion detection and outlines some of the contributions that addressed regression analysis to perform intrusion detection.
Depression is very common among patients with Alzheimer's while identifying depression in patients with Alzheimer's can be difficult, since dementia can cause some of the same symptoms. The related work in deep learning and machine learning proposed classification models that assist in detecting depression. However, classifying Alzheimer patients into depressive and non-depressive is not an easy task. Therefore, the objective of this research paper is to establish a starting point to use Artificial Neural Networks (ANN) to classify Alzheimer patients into depressive and non-depressive using speech analysis. The research paper proposes an analysis of the performance rates (accuracy, recall, precision) for ANN. The analysis performs three experiments and compare the performance rates among selected audio features. Our classification model shows promising classification results: the classification accuracy is ranged between 72.5% and 77.1%. This result provides a positive indication that ANN can assist the medical communities in future research. This could be accomplished by developing the feature extraction process, choosing the appropriate data and audio features, and developing the classification methods.
The mechanism underlying the complex behavior of major or clinical depression has been hypothesized as a dynamical system. This study presents an analysis of mood self-assessment in major depression using the method of fuzzy cross recurrence plots and tensor decomposition. The relationships between positive and negative mental states can be visualized and quantified using the proposed approach. The discovery of cross recurrences or dynamic correlations of the two opposite mental states can be helpful for targeted therapy in the management of personalized depression either under or without antidepressant drugs.
Recommender systems are used to generate meaningful recommendations to users based on their preferences, which will be determined following several approaches. This work targets Arab readers by providing accurate and reliable results that match their needs and desirability. Eventually, it will enhance the reading experience for any Arab readers. The main approach is to filter the recommendations, and this can be achieved either by Content-Based filtering or by Collaborative Filtering. The collaborative filtering techniques presented in this paper compute the similarity matrix between items and users' ratings, and then evaluate the recommendations for users. The techniques cover User-Based and Item-Based Collaborative Filtering, as well as Matrix Factorization through an SVD algorithm. A comparison between these techniques is presented in terms of the fitting and testing time, and accuracy. The KNN-based algorithms showed better performance than the matrix factorization method with respect to fitting and testing time. However, the matrix factorization (SVD) algorithm had the best results in terms of accuracy.
A cyber-attack is precautious manipulation of computer systems and networks using malware to conciliate data or restrict processes or operations. These types of attacks are vastly growing over the years. This increase in structure and complexity calls for advanced innovation in defensive strategies and detection. Traditional approaches for detecting cyber-attacks suffer from low efficiency, especially with the high demands of increasing security threats. With the substitutional increase of computational power, machine learning and deep learning methods are considered significant solutions for defending and detecting those threats or attacks. In this paper, we performed a comparative analysis of IoT cyberattack detection methods. We utilized six different algorithms including, Random Forest, Logistic Regression, SVM, NB, KNN, and MLP. Each model is evaluated using precision, recall, F-score, and ROC.
The paper is about the detection of unauthenticated news using Machine-learning methods with different algorithms. There is lot of scope to check the reality of the news received from various sources like websites, blogs, e-content. To identify the fake news, there is a need of some application in real time. Many methods were proposed earlier to observe fake news such as style-based, propagation-based and user-based. Automatic fake news detection application can be generated using natural language processing, information retrieval techniques, as well as graph theory. Language modeling is used to predict the missing or next word in a sentence based on the context. It is believed that mainstream media platforms are publishing fake news to grasp the attention of readers; most likely, it is done to increase the number of visitors on that particular page so that with an increasing number of visitors the page could claim more advertisement. This paper proposes an efficient method to detect fake news with better accuracy by using the available data set to detect the news is FAKE or REAL. Various methods are used for collecting the data and the data mining techniques are applied to clean and visualize it. Data mining helps to differentiate between the qualities of data depending upon its properties. The performance of detecting news only from the body of news is not sufficient but also social engagements should be considered. The objective of the work is to provide end-users with a robust solution so that they can figure out phishy and misguiding information. This technique combines the title and the body of the news to predict fake news more efficiently. The application is concerned with finding a result that could be used to identify fake news to help users.
In this paper, we present the main ideas behind the development of a system that can be used to deal with meteorological big data. In particular, the system captures data online and downloads it locally onto a MongoDB database. After that, the user can create a particular database and corresponding minable views for analysis. The results provided by the systems are predictive models with the ability to predict some weather-related variables, such as temperature and rainfall. The system has been validated from a triple perspective (usability, experts’ validation, and performance assessment), obtaining satisfactory results. This paper aims to be a brief guide for authors who intend to developed similar systems either in the meteorological field or other domains generating big amounts of data.
Temporal transactional databases are transactional databases which store data in a temporal aspect. Usage of similarity of measures in temporal data mining tasks have gained significant importance to retrieve information and interesting patterns in data. It is always crucial to understand and decide what similarity measure we should use while performing a data mining task and this is always driven by the actual data and nature of the temporal data sets. The main objective of this research is to perform a detailed survey of the various similarity measures used in the temporal data mining in recent research contributions. This paper also provides insights on how these similarity measures are used in the Temporal association rule mining algorithms based on the works carried out in the literature.
Graffiti is an element of graphic expression that manifests different states of the human being. However, for many governments worldwide, it has been an element of discord between them and the communities that express themselves through graffitis. This article proposes identifying graffiti and concentration zones through Computer Vision and object detection and localization to support public policy management in smart cities. ASUM-DM methodology is used to achieve the aim. Initially, the current problems faced by municipal governments in the management of public graffiti policy are identified. Then available datasets of images from Google Street View (GSV) and other acquired datasets are identified for the case study carried out in the city of Medellín (Colombia) and border municipalities. A training dataset of 1,395 images and a production dataset of 71,100 panoramas is placed on strictly using the experimental method of the division of training data, validation, and a production sample, to make a correct estimation of the generalization error. As a result of the training process, we obtained an Average Precision of 69,14%, which presented a high precision Tag of 89.23%, and low precision of 59.13% in Mural. Finally, it is possible to build heat maps of graffiti concentration areas that could guide rulers to create or improve public policies related to graffiti expression.
Challenging annotated video datasets are in huge demand for the researchers and embedded industrials to learn and build an artificial intelligence for detecting, localizing and classifying the objects of interest aimed at various applications under pattern recognition and computer vision domain. It is very significant to produce those annotated sets to the respective communal. This paper focuses on text as annotated data in video for detection, localization, tracking and classification to solve several optical character recognition (OCR) based problems. Text is very essential in understanding the nature of the video because of diverse applications which are in renowned today like video retrieval and searching, driverless cars, industrial goods automation, geocoding and many more. Hence, it is important to understand how to create, prepare and load datasets to make ready for the machine to learn and understand. First, we have applied bilateral filter to preserve the edge information. Then, rotational gradient approach is proposed to detect the text in variable viewpoints. Later, the combination of morphology and contours has applied to generate blobs with bounding box around the detected regions by eradicating quasi text areas. The simulation results have shown better performance than traditional techniques with better detection rate on ICDAR Robust Reading Competition on Text in Video 2013-15 datasets.
The proposed model for organizing blended and distance learning involves the creation of an individual learning path, which makes it flexible. The learning model is represented using an ontological model, and the decision rules for the model are logical rules. This training model is used to train teachers in digital literacy and distance learning in the context of Covid 19
Distributed Denial of Service (DDoS) attacks in the cloud environment are not as simple as the same attacks which occur in the traditional physical network environment. Not only one single attack is affecting the cloud environment, where as there are multiple sources to affect the environment. DDoS attacks can be detected using the existing machine learning techniques such as neural classifiers. This paper discusses on the survey carried out on DDoS attacks in the cloud environment. Using Machine learning techniques results to detection of higher false positive rates. Some of the widely used methods are ANN, SVM, kNN, J48, Feature rank and Feature selection methods to detect DDoS attacks in the cloud environment. This paper reviews various studies related to detection of network attacks in network and cloud environments.
