rss_2.0Applied Computer Systems FeedSciendo RSS Feed for Applied Computer Systems Computer Systems Feed Chatbot for Tourist Recommendations: A Case Study in Vietnam<abstract> <title style='display:none'>Abstract</title> <p>Living standards are rising due to a more developed society, and recreation, particularly tourism, is becoming more critical. Expanding the tourist industry is one of the most significant concerns in economic growth. Tourism revenue has helped increase residents’ income, leading to socio-economic development. In recent years, emerging Vietnamese tourism spots like Hon Son, Sapa, Hue, Phu Quoc in Vietnam, and others have consistently drawn travellers to visit and experience through social networking platforms. Tourism potential is tremendous, but foreign visitors’ information about tourist destinations still needs to be improved. This work proposes an approach to integrating machine learning algorithms into an information system to consult tourism traveling. Machine learning algorithms can classify question topics, predict user intent, and predict conversation scenarios to give appropriate responses. Our method is evaluated on the dataset, including 7319 samples on 11 topics collected from the TWCS dataset, using three algorithms: Bag of Words, BERT, and RoBERTa. BERT achieved the highest performance among the surveyed algorithms with 90 % in accuracy and 90.1 % in F1-Score. From the trained model, the team built a mobile application on Android to deploy the chatbot application with the Flutter framework based on Dart, an object-oriented programming language developed by Google using the concept of containers. The system’s functionality serves two primary user groups: administrators and application users. Administrators can utilize the application’s primary functions to manage content set up, and train a chatbot. Users can access information about locations, read location articles, check hotel prices, and use chatbots to find answers to their location-related questions. Administrators can also train the chatbot model to expand its knowledge.</p> </abstract>ARTICLEtrue Rigid Image Registration by Combined Local Features and Genetic Algorithms<abstract> <title style='display:none'>Abstract</title> <p>Image registration is an essential pre-processing step required for many image processing applications such as medical imaging and computer vision. The aim is to geometrically align two or more images of the same scene by establishing a mapping that relies on each point from one image to its corresponding point of another image. Scale invariant feature transform (SIFT) and speeded up robust features (SURF) are well-liked local features descriptors that have been extensively utilised for feature-based image registration due to their inherent properties such as invariance, changes in illumination, and noise. Moreover, the task of registration can be viewed as an optimization problem that can be solved by applying genetic algorithms (GAs). This paper presents an efficient feature image registration method based on combined local features and GAs. Firstly, the procedure consists of extracting the local features from the images by combining SIFT and SURF algorithms and matching them to refine the feature set data. Therefore, an adaptive GA based on fitness sharing and elitism techniques is employed to find the optimal rigid transformation parameters that best align the feature points by minimizing a distance metric. The suggested method is applied for registering medical images and the obtained results are significant compared to other feature-based approaches with reasonable computation time.</p> </abstract>ARTICLEtrue An Energy-Harvested Transmitter-Initiated MAC Protocol for Wireless Sensor Networks<abstract> <title style='display:none'>Abstract</title> <p>The paper proposes an Energy-Harvested Transmitter-Initiated MAC Protocol for WSNs (ETI-MAC). ETIMAC takes advantage of the benefits of transmitter-initiated schemes and employs the low power listening (LPL) method with small preamble messages so that each sensor node in the network can predict its next sleep duration based on the harvested energy rate value, thereby lowering the duty cycle by making use of its accumulating residual energy. The simulation results show that the proposed protocol outperforms compared to the old PSEHWSN scheme.</p> </abstract>ARTICLEtrue Analysis Based on Urdu Reviews Using Hybrid Deep Learning Models<abstract> <title style='display:none'>Abstract</title> <p>Worldwide websites publish enormous amounts of text, audio, and video content every day. This valuable information allows for the assessment of regional trends and general public opinion. Based on consumers’ online behavioural habits, businesses are showing them their chosen ads. It is difficult to carefully analyse these raw data to find valuable trends, especially for a language with limited resources like Urdu. There have not been many studies or efforts to create language resources for the Urdu language and analyse people’s sentiment, even though there are more than 169 million Urdu speakers in the world and a sizable amount of Urdu data is generated on various social media platforms every day. However, there has been relatively little research on sentiment analysis in Urdu. Researchers have primarily performed studies in English and Chinese. In response to this gap, we suggest an emotion analyser for Urdu, the primary language of Asia, in this research study. In this paper, we propose to assess sentiment in Urdu review texts by integrating a bidirectional long short-term memory (BiLSTM) model with a convolutional neural network (CNN). We contrast the CNN, LSTM, BiLSTM, and CNN-LSTM models with the CNN-BiLSTM model. With an accuracy rate of 0.99 %, the CNN-BiLSTM model performed better than the other models in an initial investigation.</p> </abstract>ARTICLEtrue Graphical User Interface for CBIR Framework<abstract> <title style='display:none'>Abstract</title> <p>Content-based image retrieval system (CBIR) is a well-known and widely used system for image retrieval. Most of the current CBIR systems are either command-based or specific to applications. However, due to the availability of a good computing facility, a graphical way of retrieving images may prove to be very useful for both industrial and research purposes. This paper proposes a generic and user-friendly graphical user interface (GUI) for CBIR framework. With the proposed GUI, any user with or without knowledge of CBIR can operate and retrieve images of their choice among a huge number of images. The GUI gives a vast range of facilities for selecting options. The proposed GUI is implemented and verified on a well-known image database.</p> </abstract>ARTICLEtrue English – Punjabi Aligned Parallel Corpora of Nouns from Comparable Corpora<abstract> <title style='display:none'>Abstract</title> <p>Comparable corpora are the right resources for extracting parallel data due to their abundant availability. It is of great importance where parallel data are scarce. In this study, the focus is placed on building of parallel data for Punjabi and English language pair. The raw data were collected from web contents of “Mann Ki Baat”, which is a collection of textual speeches of Prime Minister of India Mr. Narendra Modi broadcasted every last Sunday of the month. Data were cleaned and pre-processed using a natural language toolkit. An alignment model using BERT was built that aligned two textual files on a sentence level. Furthermore, extraction of noun forms with the help of NLTK library in Python programming was performed. The noun aligned dataset was built for English-Punjabi language pair and made available at Mendeley data repository.</p> </abstract>ARTICLEtrue Code Features and their Dependencies: An Aggregative Statistical Analysis on Open-Source Java Software Systems<abstract> <title style='display:none'>Abstract</title> <p>Source code constitutes the static and human-readable component of a software system. It comprises an array of artifacts and features that collectively execute a specific set of tasks. Coding behaviours and patterns are formulated through the orchestrated utilization of distinct features in a specified sequence, fostering inter-dependencies among these features. This study seeks to explore into the presence of specific coding behaviours and patterns within Java, which could potentially unveil the extent to which developers endeavour to leverage the facilities and services that exist in the programming language aggregatively. In pursuit of investigating behaviours and patterns, 436 open-source Java projects are selected, each having more than 150 Java files (Classes and Interfaces), in a semi-randomized manner. For every project, 39 features have been chosen, and the frequency of each individual feature has been independently assessed. By employing linear regression, the interrelationships among all features across the complete array of projects are scrutinized. This analysis intends to uncover the manifestation of distinct coding behaviours and patterns. Based on the selected features, preliminary findings suggest a notable collective incorporation of diverse coding behaviours among programmers, encompassing Encapsulation and Polymorphism. The findings also point to a distinct preference for using a specific commenting mechanism, JavaDoc, and the potential existence of Code-Clone and dead code. Overall, the results indicate a clear tendency among programmers to strongly adhere to the fundamental principles of Object -Oriented programming. However, certain less obvious attributes of object-oriented languages appear to receive relatively less attention from programmers.</p> </abstract>ARTICLEtrue Improved FakeBERT for Fake News Detection<abstract> <title style='display:none'>Abstract</title> <p>In the present era of the internet and social media, the way of information dissemination has changed. However, due to rapid growth in the amount of news generated regularly and the unsupervised nature of social media, fake news turns out to be a big problem. Fake news can easily build a false positive or negative perception about a person, or an event. Fake news was also used as a tool by propagandists during the Coronavirus (COVID-19) pandemic. Thus, there is a need to use technology to tag fake news and prevent its dissemination. Previously, different algorithms were designed to detect fake news but without considering the semantic meaning and long sentence dependence. This research work proposes a new approach to the detection of fake news in the context of COVID-19. The suggested approach uses a combination of Bidirectional Encoder Representations from Transformers (BERT) for extracting context meaning from sentences, SVM for pattern identification to detect fake news in a better way from the COVID-19 dataset, and an evolutionary algorithm called Non-dominated Sorting Genetic Algorithm II (NSGA-II) to distribute text for Support Vector Machine (SVM) classification. The suggested approach improves accuracy by 5.2 % by removing a certain amount of ambiguity from sentences.</p> </abstract>ARTICLEtrue of the Traveling Salesman Problem by Clustering its Nodes and Finding the Best Route Passing through the Centroids<abstract> <title style='display:none'>Abstract</title> <p>A method of heuristically solving large and extremely large traveling salesman problems is suggested. The solver is a specific genetic algorithm producing approximately shortest routes the fastest of known heuristics without losing much in accuracy. The method consists in parallelizing the problem by clustering its nodes and finding the best route passing through the centroids of the clusters. The open-loop subroutes of the clusters are connected via specific nodes. These specific nodes referred to as connectors are determined as those for which the distance to the depot is maximal and the distance to the cluster of the following subproblem is minimal. Thus, a bunch of smaller open-loop problems is solved instead of solving the whole (closed loop) problem. Extremely large problems should be clustered manually by imposing a mesh of rotated square cells. In this case, the connectors should be determined manually as well. A connector can also be approximated by a node which is the closest to the line connecting the centroids of the two clusters. The suggested parallelization can produce a very significant speedup depending on how many processor cores are simultaneously available. The factual speedup by the parallelization depends on the availability of processor cores, memory, and the processor clock frequency. The efficiency of the parallelization is maintained for a few hundred to a few million nodes by any number of clusters being less than the size of the average cluster.</p> </abstract>ARTICLEtrue Model for the Specification of Multi-view Point Ontology<abstract> <title style='display:none'>Abstract</title> <p>In this paper, we propose a new approach, based on bigraphic reactive systems (BRS), to provide a formal modelling of the architectural elements of a Multi-Viewpoints ontology (MVp ontology). We introduce a formal model in which the main elements of MVp ontology find their definition in terms of bigraphic concepts by preserving their semantics. Besides, we enrich the proposed model with reaction rules in order to handle the dynamic reactions of MVp ontology. In order to confirm the applicability of our approach, we have carried out a case study using the proposed model.</p> </abstract>ARTICLEtrue Nearest Neighbour-based Index Tree: A Case Study for Instrumental Music Search<abstract> <title style='display:none'>Abstract</title> <p>Many people are interested in instrumental music. They may have one piece of song, but it is a challenge to seek the song because they do not have lyrics to describe for a text-based search engine. This study leverages the Approximate Nearest Neighbours to preprocess the instrumental songs and extract the characteristics of the track in the repository using the Mel frequency cepstral coefficients (MFCC) characteristic extraction. Our method digitizes the track, extracts the track characteristics, and builds the index tree with different lengths of each MFCC and dimension number of vectors. We collected songs played with various instruments for the experiments. Our result on 100 pieces of various songs in different lengths, with a sampling rate of 16000 and a length of each MFCC of 13, gives the best results, where accuracy on the Top 1 is 36 %, Top 5 is 4 %, and Top 10 is 44 %. We expect this work to provide useful tools to develop digital music e-commerce systems.</p> </abstract>ARTICLEtrue of the -Means Algorithm for Partitioning Large Datasets of Flat Points by a Preliminary Partition and Selecting Initial Centroids<abstract> <title style='display:none'>Abstract</title> <p>A problem of partitioning large datasets of flat points is considered. Known as the centroid-based clustering problem, it is mainly addressed by the <italic>k</italic>-means algorithm and its modifications. As the <italic>k</italic>-means performance becomes poorer on large datasets, including the dataset shape stretching, the goal is to study a possibility of improving the centroid-based clustering for such cases. It is quite noticeable on non-sparse datasets that the resulting clusters produced by <italic>k</italic>-means resemble beehive honeycomb. It is natural for rectangular-shaped datasets because the hexagonal cells make efficient use of space owing to which the sum of the within-cluster squared Euclidean distances to the centroids is approximated to its minimum. Therefore, the lattices of rectangular and hexagonal clusters, consisting of stretched rectangles and regular hexagons, are suggested to be successively applied. Then the initial centroids are calculated by averaging within respective hexagons. These centroids are used as initial seeds to start the <italic>k</italic>-means algorithm. This ensures faster and more accurate convergence, where at least the expected speedup is 1.7 to 2.1 times by a 0.7 to 0.9 % accuracy gain. The lattice of rectangular clusters applied first makes rather rough but effective partition allowing to optionally run further clustering on parallel processor cores. The lattice of hexagonal clusters applied to every rectangle allows obtaining initial centroids very quickly. Such centroids are far closer to the solution than the initial centroids in the <italic>k</italic>-means++ algorithm. Another approach to the <italic>k</italic>-means update, where initial centroids are selected separately within every rectangle hexagons, can be used as well. It is faster than selecting initial centroids across all hexagons but is less accurate. The speedup is 9 to 11 times by a possible accuracy loss of 0.3 %. However, this approach may outperform the <italic>k</italic>-means algorithm. The speedup increases as both the lattices become denser and the dataset becomes larger reaching 30 to 50 times.</p> </abstract>ARTICLEtrue Mobile User Profiling for Maximum Performance<abstract> <title style='display:none'>Abstract</title> <p>The use of smartphones and their applications is expanding rapidly, thereby increasing the demand of computational power and other hardware resources of the smartphones. On the other hand, these small devices can have limited resources of computation power, battery backup, RAM memory, and storage space due to their small size. These devices need to reconcile resource hungry applications. This research focuses on solving issues of power and efficiency of smart devices by adapting intelligently to mobile usage by profiling the user intelligently. Our designed architecture makes a smartphone smarter by intelligently utilizing its resources to increase the battery life. Our developed application makes profiles of the applications usage at different time intervals. These stored usage profiles are utilized to make intelligent resource allocation for next time interval. We implemented and evaluated the profiling scheme for different brands of android smartphone. We implemented our approach with Naive Bayes and Decision Tree for performance and compared it with conventional approach. The results show that the proposed approach based on decision trees saves 31 % CPU and 60 % of RAM usage as compared to the conventional approach.</p> </abstract>ARTICLEtrue and 3D Visualization of Human Body Parts and Bone Areas Using CT Images<abstract> <title style='display:none'>Abstract</title> <p>The advent of medical imaging significantly assisted in disease diagnosis and treatment. This study introduces to a framework for detecting several human body parts in Computerised Tomography (CT) images formatted in DICOM files. In addition, the method can highlight the bone areas inside CT images and transform 2D slices into a visual 3D model to illustrate the structure of human body parts. Firstly, we leveraged shallow convolutional Neural Networks to classify body parts and detect bone areas in each part. Then, Grad-CAM was applied to highlight the bone areas. Finally, Insight and Visualization libraries were utilized to visualize slides in 3D of a body part. As a result, the classifiers achieved 98 % in F1-score in the classification of human body parts on a CT image dataset, including 1234 slides capturing body parts from a woman for the training phase and 1245 images from a male for testing. In addition, distinguishing between bone and non-bone images can reach 97 % in F1-score on the dataset generated by setting a threshold value to reveal bone areas in CT images. Moreover, the Grad-CAM-based approach can provide clear, accurate visualizations with segmented bones in the image. Also, we successfully converted 2D slice images of a body part into a lively 3D model that provided a more intuitive view from any angle. The proposed approach is expected to provide an interesting visual tool for supporting doctors in medical image-based disease diagnosis.</p> </abstract>ARTICLEtrue Deep SLAM-CNN Assisted Underwater SLAM<abstract> <title style='display:none'>Abstract</title> <p>Underwater simultaneous localization and mapping (SLAM) poses significant challenges for modern visual SLAM systems. The integration of deep learning networks within computer vision offers promising potential for addressing these difficulties. Our research draws inspiration from deep learning approaches applied to interest point detection and matching, single image depth prediction and underwater image enhancement. In response, we propose 3D-Net, a deep learning-assisted network designed to tackle these three tasks simultaneously. The network consists of three branches, each serving a distinct purpose: interest point detection, descriptor generation, and depth prediction. The interest point detector and descriptor generator can effectively serve as a front end for a classical SLAM system. The predicted depth information is akin to a virtual depth camera, opening up possibilities for various applications. We provide quantitative and qualitative evaluations to illustrate some of these potential uses. The network was trained in in several steps, using in-air datasets and followed by generated underwater datasets. Further, the network is integrated into feature-based SALM systems ORBSLAM2 and ORBSSLAM3, providing a comprehensive assessment of its effectiveness for underwater navigation.</p> </abstract>ARTICLEtrue Media: An Exploratory Study of Information, Misinformation, Disinformation, and Malinformation<abstract> <title style='display:none'>Abstract</title> <p>The widespread use of social media all around the globe has affected the way of life in all aspects, not only for individuals but for businesses as well. Businesses share their upcoming events, reveal their products, and advertise to their potential customers, where individuals use social media to stay connected with their social circles, get updates and news from social media pages of news agencies, and update their information regarding the latest activities, businesses, economy, events, politics, trends, and about their area of interest. According to Statista, there were 4.59 billion users of social media worldwide in 2022 and expected to grow up to 5.85 billion in the year 2027. With its massive user base, social media does not only generate useful information for businesses and individuals, but at the same time, it also creates an abundance of misinformation and disinformation as well as malinformation to acquire social-political or business agendas. Individuals tend to share social media posts without checking the authenticity of the information they are sharing, which results in posts having misinformation, disinformation, or malinformation becoming viral around the world in a matter of minutes. Identifying misinformation, disinformation, and malinformation has become a prominent problem associated with social media.</p> </abstract>ARTICLEtrue Genetic-Based Wolf Optimization for Load Balancing in Cloud Computing<abstract> <title style='display:none'>Abstract</title> <p>Cloud remains an active and dominant player in the field of information technology. Hence, to meet the rapidly growing requirement of computational processes and storage resources, the cloud provider deploys efficient data centres globally that comprise thousands of IT servers. Because of tremendous energy and resource utilization, a reliable cloud platform has to be necessarily optimized. Effective load balancing is a great option to overcome these issues. However, loading balancing difficulties, such as increased computational complexity, the chance of losing the client data during task rescheduling, and consuming huge memory of the host, and new VM (Virtual Machine), need appropriate optimization. Hence, the study aims to create a newly developed IG-WA (Inquisitive Genetic–Wolf Optimization) framework that meritoriously detects the optimized virtual machine in an environment. For this purpose, the system utilises the GWO (Grey Wolf Optimization) method with an evolutionary mechanism for achieving a proper compromise between exploitation and exploration, thereby accelerating the convergence and achieving optimized accuracy. Furthermore, the fitness function evaluated with an inquisitive genetic algorithm adds value to the overall efficacy. Performance evaluation brings forward the outperformance of the proposed IGWO system in terms of energy consumption, execution time and cost, makespan, CPU utilization, and memory utilization. Further, the system attains more comprehensive and better results when compared to the state of art methods.</p> </abstract>ARTICLEtrue Analysis of Supervised and Unsupervised Machine Learning Algorithms with Aspect-Based Sentiment Analysis<abstract> <title style='display:none'>Abstract</title> <p>Machine learning based sentiment analysis is an interdisciplinary approach in opinion mining, particularly in the field of media and communication research. In spite of their different backgrounds, researchers have collaborated to test, train and again retest the machine learning approach to collect, analyse and withdraw a meaningful insight from large datasets. This research classifies the texts of micro-blog (tweets) into positive and negative responses about a particular phenomenon. The study also demonstrates the process of compilation of corpus for review of sentiments, cleaning the body of text to make it a meaningful text, find people’s emotions about it, and interpret the findings. Till date the public sentiment after abrogation of Article 370 has not been studied, which adds the novelty to this scientific study. This study includes the dataset collection from Twitter that comprises 66.7 % of positive tweets and 34.3 % of negative tweets of the people about the abrogation of Article 370. Experimental testing reveals that the proposed methodology is much more effective than the previously proposed methodology. This study focuses on comparison of unsupervised lexicon-based models (TextBlob, AFINN, Vader Sentiment) and supervised machine learning models (KNN, SVM, Random Forest and Naïve Bayes) for sentiment analysis. This is the first study with cyber public opinion over the abrogation of Article 370. Twitter data of more than 2 lakh tweets were collected by the authors. After cleaning, 29732 tweets were selected for analysis. As per the results among supervised learning, Random Forest performs the best, whereas among unsupervised learning TextBlob achieves the highest accuracy of 99 % and 88 %, respectively. Performance parameters of the proposed supervised machine learning models also surpass the result of the recent study performed in 2023 for sentiment analysis.</p> </abstract>ARTICLEtrue Biometric System Based on the Fusion in Score of Fingerprint and Online Handwritten Signature<abstract> <title style='display:none'>Abstract</title> <p>Multimodal biometrics is the technique of using multiple modalities on a single system. This allows us to overcome the limitations of unimodal systems, such as the inability to acquire data from certain individuals or intentional fraud, while improving recognition performance. In this paper, a study of score normalization and its impact on the performance of the system is performed. The fusion of scores requires prior normalisation before applying a weighted sum fusion that separates impostor and genuine scores into a common interval with close ranges. The experiments were carried out on three biometric databases. The results show that the proposed strategy performs very encouragingly, especially in combination with Empirical Modal Decomposition (EMD). The proposed fusion system shows good performance. The best result is obtained by merging the globality online signature and fingerprint where an EER of 1.69 % is obtained by normalizing the scores according to the Min-Max method.</p> </abstract>ARTICLEtrue Approach for Sentiment Analysis Using Stack of Neural Network with Lexicon Based Padding and Attention Mechanism<abstract> <title style='display:none'>Abstract</title> <p>Sentiment analysis (SA) has been an important focus of study in the fields of computational linguistics and data analysis for a decade. Recently, promising results have been achieved when applying DNN models to sentiment analysis tasks. Long short-term memory (LSTM) models, as well as its derivatives like gated recurrent unit (GRU), are becoming increasingly popular in neural architecture used for sentiment analysis. Using these models in the feature extraction layer of a DNN results in a high dimensional feature space, despite the fact that the models can handle sequences of arbitrary length. Another problem with these models is that they weight each feature equally. Natural language processing (NLP) makes use of word embeddings created with word2vec. For many NLP jobs, deep neural networks have become the method of choice. Traditional deep networks are not dependable in storing contextual information, so dealing with sequential data like text and sound was a nightmare for such networks. This research proposes multichannel word embedding and employing stack of neural networks with lexicon-based padding and attention mechanism (MCSNNLA) method for SA. Using convolution neural network (CNN), Bi-LSTM, and the attention process in mind, this approach to sentiment analysis is described. One embedding layer, two convolution layers with max-pooling, one LSTM layer, and two fully connected (FC) layers make up the proposed technique, which is tailored for sentence-level SA. To address the shortcomings of prior SA models for product reviews, the MCSNNLA model integrates the aforementioned sentiment lexicon with deep learning technologies. The MCSNNLA model combines the strengths of emotion lexicons with those of deep learning. To begin, the reviews are processed with the sentiment lexicon in order to enhance the sentiment features. The experimental findings show that the model has the potential to greatly improve text SA performance.</p> </abstract>ARTICLEtrue