17 research outputs found
A Multiagent-based Framework for Integrating Biological Data
No full textBiological data has been rapidly increasing in volume in different Web data sources. To query multiple data sources manually on the internet is time consuming for biologists. Therefore, systems and tools that facilitate searching multiple biological data sources are needed. Traditional approaches to build distributed or federated systems do not scale well to the large, diverse, and the growing number of biological data sources. Internet search engines allow users to search through large numbers of data sources, but provide very limited capabilities for locating, combining, processing, and organizing information. A promising approach to this problem is to provide access to the large number of biological data sources through a multiagent-based framework where a set of agents can cooperate with each other to retrieve relevant information from different biological Web databases. The proposed system uses a mediator-based integration approach with domain ontology, which uses as a global schema. In this paper we propose a multiagent-based framework that responds to biological queries according to its biological domain ontology.</p
Fog-Based CDN Framework for Minimizing Latency of Web Services Using Fog-Based HTTP Browser
No full textCloud computing has been a dominant computing paradigm for many years. It provides applications with computing, storage, and networking capabilities. Furthermore, it enhances the scalability and quality of service (QoS) of applications and offers the better utilization of resources. Recently, these advantages of cloud computing have deteriorated in quality. Cloud services have been affected in terms of latency and QoS due to the high streams of data produced by many Internet of Things (IoT) devices, smart machines, and other computing devices joining the network, which in turn affects network capabilities. Content delivery networks (CDNs) previously provided a partial solution for content retrieval, availability, and resource download time. CDNs rely on the geographic distribution of cloud servers to provide better content reachability. CDNs are perceived as a network layer near cloud data centers. Recently, CDNs began to perceive the same degradations of QoS due to the same factors. Fog computing fills the gap between cloud services and consumers by bringing cloud capabilities close to end devices. Fog computing is perceived as another network layer near end devices. The adoption of the CDN model in fog computing is a promising approach to providing better QoS and latency for cloud services. Therefore, a fog-based CDN framework capable of reducing the load time of web services was proposed in this paper. To evaluate our proposed framework and provide a complete set of tools for its use, a fog-based browser was developed. We showed that our proposed fog-based CDN framework improved the load time of web pages compared to the results attained through the use of the traditional CDN. Different experiments were conducted with a simple network topology against six websites with different content sizes along with a different number of fog nodes at different network distances. The results of these experiments show that with a fog-based CDN framework offloading autonomy, latency can be reduced by 85% and enhance the user experience of websites
Factors Influencing Investment Decisions in Financial Investment Companies
No full textFor making the most favorable financial decisions possible, it is essential to have an understanding of aspects and the factors which can play a role in the decision-making. In contrast to previous research on the subject, which has only examined a single factor in making investment decisions, our study takes a more holistic approach by looking at several factors. The purpose of this study was to discover the elements that influence investment decisions made by financial organizations that are listed on Iraqi stock exchanges (ISX). The research was carried out on the six companies that made up the study’s sample size. For the purpose of data collection, the researcher utilized a structured questionnaire that was delivered to the respondents in an individual capacity. The questionnaire contained eight items. The factors of the questionnaire were analyzed with respect to normal distribution, the problem of linear multiplicity, the validity of the questionnaire in terms of content and appearance, the stability of the questionnaire by the split-half method, and the test and re-test method. In addition, the research hypotheses were tested on both the independent variables and the dependent variables. We calculated the mean, standard deviation, weight percentile, and coefficient of variance from the collected data. The significance of the connection between the dimensions of the decision-making factors was clarified through the use of Spearman’s correlation coefficient and the t test. We concluded that in the last step of the proposed model there is an increase in coefficients of determination and it reaches a value of (0.98), which is a very excellent and almost complete interpretation of the impact of dimensions extracted in the model and their impact on investment decision. As is noted, a slight decline in the value of the regression coefficient for all variables occurred, and also we noticed that the signs for the coefficients for the five variables are positive, meaning that they reflect the extent of the direct effect of those variables in making the investment decision. The response rate for the questionnaire was 97.7%
Arabic speech recognition using end‐to‐end deep learning
No full textAbstract Arabic automatic speech recognition (ASR) methods with diacritics have the ability to be integrated with other systems better than Arabic ASR methods without diacritics. In this work, the application of state‐of‐the‐art end‐to‐end deep learning approaches is investigated to build a robust diacritised Arabic ASR. These approaches are based on the Mel‐Frequency Cepstral Coefficients and the log Mel‐Scale Filter Bank energies as acoustic features. To the best of our knowledge, end‐to‐end deep learning approach has not been used in the task of diacritised Arabic automatic speech recognition. To fill this gap, this work presents a new CTC‐based ASR, CNN‐LSTM, and an attention‐based end‐to‐end approach for improving diacritisedArabic ASR. In addition, a word‐based language model is employed to achieve better results. The end‐to‐end approaches applied in this work are based on state‐of‐the‐art frameworks, namely ESPnet and Espresso. Training and testing of these frameworks are performed based on the Standard Arabic Single Speaker Corpus (SASSC), which contains 7 h of modern standard Arabic speech. Experimental results show that the CNN‐LSTM with an attention framework outperforms conventional ASR and the Joint CTC‐attention ASR framework in the task of Arabic speech recognition. The CNN‐LSTM with an attention framework could achieve a word error rate better than conventional ASR and the Joint CTC‐attention ASR by 5.24% and 2.62%, respectively
Forecasting Financial Investment Firms’ Insolvencies Empowered with Enhanced Predictive Modeling
No full textIn the realm of financial decision-making, it is crucial to consider multiple factors, among which lies the pivotal concern of a firm’s potential insolvency. Numerous insolvency prediction models utilize machine learning techniques try to solve this critical aspect. This paper aims to assess the financial performance of financial investment firms listed on the Iraq Stock Exchange (ISX) from 2012 to 2022. A Multi-Layer Perceptron predicting model with a parameter optimizer is proposed integrating an additional feature selection process. For this latter process, three methods are proposed and compared: Principal Component Analysis, correlation coefficient, and Particle Swarm Optimization. Through the fusion of financial ratios with machine learning, our model exhibits improved forecast accuracy and timeliness in predicting firms’ insolvency. The highest accuracy model is the integrated MLP + PCA model, at 98.7%. The other models, MLP + PSO and MLP + CC, also exhibit strong performance, with 0.3% and 1.1% less accuracy, respectively, compared to the first model, indicating that the first model serves as a powerful predictive approach
A Hybrid Feature Selection Approach for Arabic Documents Classification
No full textText Categorization (classification) is the process of classifying documents into a predefined set of categories based on their content. Text categorization algorithms usually represent documents as bags of words and consequently have to deal with huge number of features. Feature selection tries to find a set of relevant terms to improve both efficiency and generalization. There are two main approaches for feature selection, local and global. In Arabic text categorization it was found that using global feature selection gives higher results but may affect some documents in a way so that they do not show any terms in the set of selected features. On the other hand local feature selection is used to overcome this problem but gives lower classification rate. In this paper a hybrid approach of global and local feature selection technique is proposed and compared with both local and global feature selection techniques. Results are reported on a set of 1132 document of six different topics showing that the proposed hybrid feature selection overcome the disadvantages of both of feature selection approaches
Arabic Text Classification Using Support Vector Machines
No full textText classification (TC) is the process of classifying documents into a predefined set of categories based on their content. Arabic language is highly inflectional and derivational language which makes text mining a complex task. In this paper we applied the Support Vector Machines (SVM) model in classifying Arabic text documents. The results compared with the other traditional classifiers Bayes classifier, K-Nearest Neighbor classifier and Rocchio classifier. Two experiments used to test the different classifiers. The first uses the training set as the test set, and the second uses Leave one testing method. Experimental results performed on a set of 1132 document show that Rocchio classifier gives better results when the size of feature set is small while SVM outperform the other classifiers when the size of the feature set is large enough. Classification rate exceeds 90% when using more than 4000 feature. Leave one method gives more realistic results over the use of training set as a test set
