In response to the challenges posed by the "new normal," this study introduces WeBarangay, a website designed to serve as a communication platform and deliver crucial barangay services. The primary goal is to offer an alternative means of interaction between barangay officials and residents in the Philippines. Utilizing the Technology Acceptance Model (TAM), the study evaluates the viability and acceptance of WeBarangay, focusing on key elements: Perceived Usefulness (PU), Perceived Ease of Use (PEOU), Attitude Towards Use (ATU), and Intention to Use (ITU). The positive outcomes of the TAM evaluation affirm that users perceive WeBarangay as useful and easy to use and express a strong intention to adopt it. The Cronbach's alpha test ensures the reliability of survey responses, highlighting consistent positive attitudes across all TAM elements. With successful development and implementation, WeBarangay emerges as a promising tool to enhance communication and service delivery in barangays, contributing to more efficient and effective community engagement.
In the process of online shopping, when consumers encounter products or services with both positive and negative reviews, they experience attitudinal ambivalence, making it challenging for them to make prompt decisions. Existing studies have not effectively integrated consumers' ambivalent attitudes to accurately recommend reviews. Therefore, this paper proposes a review recommendation method that considers consumers' ambivalent attitudes. To achieve this, we first combine sentiment analysis with an ambivalence measurement formula to categorize consumers into different ambivalence groups. Next, we employ a combination of LDA and Word2vec models to extract the concerns of these distinct ambivalence groups. Finally, we utilize a Logistic regression model to match the ambivalence groups with users' basic attribute indicators and offer review recommendations tailored to address the challenge of consumers facing difficulty in making quick decisions due to attitudinal ambivalence.The example results of mafengwo platform data show that this study can divide ambivalence groups through sentiment analysis and develop differentiated recommendation strategies.
Assessing the engagement of students in online classroom is crucial to meet their learning objectives. Many machine learning and deep learning models have been proposed to handle this problem using a variety of sensors, with videos cameras being the most prominent. However, most of these approaches are not interoperable because different datasets use different labeling protocols. As a result, the classification models range from binary, multi-class to regression problems. Another problem is the lack of rigor and definition of engagement to annotate the data. In this paper, firstly we showed inconsistencies in the labeling of a popular student engagement DAiSEE dataset. Then, we re-labeled more than 7000 videos of this dataset using a methodical engagement annotation protocol, HELP, to convert it from four class to binary classification problem. Further analysis highlights issues in DAiSEE annotation in comparison to the HELP protocol. Lastly, we tested three state-of-the-art deep learning and feature-based methods and discussed their performance. Data imbalance in the newly and previously annotated data was found to be the main issue in developing predictive models.
Online learning is currently developing rapidly, but the separation between teachers and students' physical spaces leads to a lack of emotional communication, which is a huge challenge that online learning currently faces. This study aims to design a multimodal emotion recognition system for online learning, to achieve real-time monitoring and feedback on the emotional states of online learners, in order to facilitate teachers' real-time adjustment and improvement of online teaching resources. This article first reviews and evaluates relevant research in the field of online learning, and then proposes a design scheme based on a multimodal emotion model, which is implemented and validated. Finally, this article verifies the effectiveness and superiority of the system through experiments.
This study introduces the Auto-Gardener, a smart gardening system designed to enhance urban gardening practices. The Auto-Gardener incorporates an impact sensor and real-time data analysis to automate critical aspects of plant care, providing users with a user-friendly and efficient gardening assistant. The study evaluates the system's usability through the System Usability Scale (SUS) and the Post-Study System Usability Questionnaire (PSSUQ), resulting in an impressive 80.6 SUS score and a 1.84 PSSUQ Overall Mean. These scores reflect users' high satisfaction and acceptance of the Auto-Gardener. The innovative features of the Auto-Gardener contribute to achieving study objectives, showcasing the potential of technology to revolutionize urban gardening. The positive outcomes underscore the importance of user-centric design in smart gardening technology. As we continue to embrace technological advancements, the Auto-Gardener exemplifies a transformative tool that can make urban gardening more accessible, efficient, and enjoyable for users.
In the contemporary realm, web applications hold paramount significance in the lives of a vast multitude, thereby rendering web application security an imperative necessity. The inherent interconnectedness of web applications in today's digital landscape renders them an exceedingly attractive target for cybercriminals to exploit. In this paper, we shall present a comprehensive analysis of prevalent web application vulnerabilities, encompassing the likes of SQL injection, OS Command Injection, Local File Inclusion, and XML External Entities. We shall examine attack scenarios that exploit these vulnerabilities and present code snippets to facilitate development.
Sales forecasting is crucial for effective marketing, inventory management, and production planning in the tobacco industry. However, predicting cigarette sales can be challenging due to the volatility of time series data, which is influenced by various factors. Traditional prediction methods often struggle to handle complex nonlinear relationships. To address this, this paper proposes a novel approach that combines empirical mode decomposition (EMD) and long-short-term memory networks (LSTM) to accurately forecast cigarette sales. The new method utilizes a rolling mechanism and comprises three main steps. Firstly, the EMD method is employed to decompose the original sales time series into several simple and relatively regular components. Next, LSTM is applied independently to predict each of these time series components. Finally, a straightforward addition and integration method is used to aggregate the individual prediction results, yielding the final sales prediction. To evaluate the performance of the proposed method, real sales data from tobacco companies is used and five long-term sales cigarette products are selected. The prediction results are then compared with those obtained from two benchmark models, LSTM and ARIMA. The empirical analysis demonstrates that the proposed model outperforms traditional prediction models in terms of both prediction accuracy and stability. These findings corroborate that the novel methodology provides a robust and reliable instrument for forecasting cigarette sales.
In a competitive market, market segmentation is essential for building effective marketing strategies. In this study, we propose two approaches to classify entities of a market into four segments which can be interpreted in advance without arbitrariness, using two quantifiable indicators. One is "FlexBound-Seg," which flexibly constructs the boundaries of each region by solving a combinatorial optimization problem, and the other is "PercentSquare-Seg," which revises the classical method in Pareto analysis to satisfy an upper bound on the number of elements in the heavy region. By applying real data of e-shopping usage, the results showed that "FlexBound-Seg," "PercentSquare-Seg," and the classical method "Pareto-Seg" in that order, provided more accurate segmentation for heavy users, and were useful methods that could cope with fluctuations in user usage over time. In this case study, it was also found that all the methods were effective in increasing the number of users through sales campaigns, and that the extraction of heavy users during the entire period was a classification method that can also look at heavy users classified into heavy region frequently. Although there are effective cases at the current stage, further application of the method to data from other industries is an issue for the future to measure its versatility. In addition, we also need to improve the method to expand the number of indicators, such as three indicators, to enable more complex segmentation.
Quality assurance in the semiconductor industry is vital, as this ensures that products continuously meet dynamic specifications expected from newly introduced technologies. Compromising the quality assurance of products negatively affects an organization's operational efficiency, urging in-depth analyses of quality assurance procedures to initiate improvement efforts targeting the issues that lower operational efficiency. With that, this study analyzed the quality assurance of a semiconductor company through the in-line sampling (ILS) procedure to quantify its current performance and offer appropriate improvement measures. Problems observed in the current ILS process include long testing durations, the presence of reject units even after testing, and missing data entries. To solve the mentioned issues, a dashboard was developed using PowerBI, showing an overview of operation productivity. Using the dashboard, ILS data from the Structured Query Language database was organized and modeled into a Markov process. Through the Markov Analysis, it was predicted that lots tested in ILS pass 89% of the time and consume 112 minutes. As for the general ILS analysis, it was found that waiting times in the verification stages contribute to the long ILS durations. These findings prompted recommendations that include information system improvement, data entry cascade training for employees, and utilization of failure binning results as criteria in endorsing units to an ILS stage.
In the database-as-a-service (DBaaS) paradigm, in untrusted environments, the cloud service provider (CSP) may send incorrect query results to the client. The client typically uses evidence-based verification methods to check the integrity of the results locally. However, the light client may not have sufficient resources to perform the verification computation. Solutions based on the Trusted Execution Environment (TEE) use the TEE as an additional trust anchor by transferring the verification computation to be performed within the CSP. However, most existing work only considers verifiable range queries on primary key columns and performs poorly on non-primary key columns. In addition, the current solutions are less efficient in verifying query results for both PK column and non-PK column data. To address the above issues, we propose an SGX-based verifiable range query solution for the light client. Our solution consists of two parts: untrusted memory and SGX-based TEE. In the untrusted memory, we propose a novel authenticated data structure (ADS) called optimizing verifiable skip list (OVSL) to efficiently protect the integrity of primary key columns; meanwhile, we use an OVSL-based two-layer ADS to ensure the integrity of non-primary key columns. In TEE, We store the root node of the ADS as well as the lowest common ancestors (LCAs) and utilize them in TEE to perform authentication operations. Finally, the experiment proves the feasibility and effectiveness of our solution.
Vehicle detection is a key technology in intelligent transportation systems, and it determines the position and categorization of object vehicles. Image-based vehicle identification technology has advanced significantly in recent years because of the growing number of cars and the popularity of traffic monitoring information systems, and it has become a hot topic in computer vision. A vehicle classification technique based on an upgraded residual network is suggested to improve the feature extraction and identification capabilities of models for vehicle pictures in crossing settings. The activation function's location on the residual block is enhanced, and the normal convolution in the residual block is replaced with a group convolution. In the residual block, an attention mechanism is then introduced. Finally, the cross-entropy loss function is replaced with the focal loss function. The Stanford Cars public dataset is used for pre-training, and a self-built crossing vehicle dataset is used for migration learning. In both datasets, the proposed model outperforms various traditional deep learning models in terms of classification accuracy.
Fuzzy testing is one of the most popular vulnerability mining techniques recently, it plays a huge role in exploiting software security vulnerabilities and improving software security. Fuzzy testing mainly performs specific variations on the collected seeds to obtain a large number of test cases that can be used to execute the target program and trigger potential crashes in the program. However, traditional fuzzy testing generally suffers from a low level of test automation and fewer types of vulnerabilities detected. Aiming at the above problems, the application of machine learning techniques to fuzzy testing has become a hot research topic in academia. However, some recent studies still have problems, such as low edge coverage and poor generalization ability. Therefore, this paper proposes a strongly directed fuzz testing method based on attention mechanism and we name the fuzzer as AMNeuzz. AMNeuzz uses neural networks combined with attention mechanisms to construct an automatic sample generation model, which is trained so that the model learns the intrinsic formatting features of the samples, thus being able to automatically generate test samples that conform to certain syntactic specifications to quickly examine program paths that may have vulnerabilities, and this improves efficiency. In addition, the performance of the fuzzers is improved by improving Neuzz's gradient strategy. The final experimental results show that the AMNeuzz method proposed in this paper can achieve higher edge coverage than NEUZZ under the same time overhead.
Keeping fish as pets has become increasingly popular, but it can be challenging to maintain a suitable environment for Arowana fish. The water quality, feeding, temperature, and lighting conditions must all be optimized, and the aquarium must be manually monitored. To address these challenges, a proposed solution is implementing a real-time monitoring system with sensors. This system would continuously monitor parameters such as temperature and pH levels, providing owners with up-to-date information without the need for manual checks. It would also include an aeration system to ensure sufficient oxygen supply and facilitate water renewal operations to maintain water quality. To make monitoring even more convenient, an IoT-based system could be implemented, allowing users to monitor and control the aquarium remotely through a mobile application. This would allow owners to check on the aquarium from anywhere, even when they are not at home. The intelligent aquarium management system would prioritize the well-being of the fish. It would provide precise monitoring and control over feeding operations, preventing health issues caused by improper feeding. It would also reduce the manual effort required for aquarium maintenance, freeing up time for owners to enjoy their pets. Overall, the Smart AroTank system has been successfully implemented and is well-accepted by users. It has been shown to be effective in meeting user needs, providing useful information, offering a well-designed interface, and eliciting high user satisfaction.
In contemporary society, technological devices have seamlessly integrated into our daily routines, becoming indispensable components that optimize time and energy consumption. Among these technologies, Artificial Intelligence (AI) and the Internet of Things (IoT) have revolutionized various aspects of life, elevating speed, convenience, and efficiency. However, this reliance on technology has brought forth security concerns that permeate every sector, impacting the network as a whole. This study proposes a solution employing a hashing mechanism for authentication using SHA-256 and implementing the SDN controller for streamlined network operations and detection of potential threats. To address this issue, the development of a comprehensive system capable of authenticating nodes, effectively routing data packets, monitoring network behavior, and detecting malicious nodes is imperative. This is to examine critical aspects necessary to achieve a robust and secure network infrastructure. This results in effective and efficient detection, routing and maintainability of the network while keeping the throughput low avoiding delay in performance and adaptive to most situations.
This study presents the development and testing of a precise cockfighting training aid leveraging a modified impact sensor microphone. The research aimed to enhance cockfighting practices by providing trainers with a reliable tool to optimize rooster capabilities. The device, integrated with the SenSabong App, accurately recorded roosters' hits, distinguishing between fight-related and incidental impacts. Rigorous testing confirmed the device's reliability and effectiveness in improving rooster performance with recommended training procedures. The findings contribute to ethical and effective cockfighting training, representing a crucial step toward advancing the sport. Recommendations include integrating wireless connectivity for enhanced user experiences and incorporating gyroscope and accelerometer technologies for precise and real-time training feedback.
In an era where digital imagery is proliferating at an unprecedented rate on social media platforms, organizing and understanding these visual resources has become a challenge of significant importance. This study introduces a novel approach to image clustering, leveraging the synergy between multimodal image-to-text transformation and advanced topic modeling techniques. Utilizing the Large Language and Vision Assistant (LLaVA) to generate detailed textual descriptions and BERTopic for subsequent clustering, we have analyzed and categorized images from Flickr into eleven distinct topics. Our methodology transcends traditional visual feature-based clustering by incorporating the contextual richness of textual data, thereby aiming to enhance the granularity and relevance of the clustering results. The performance of this approach was quantitatively evaluated using precision, recall, and F1 scores, revealing overall promising results with some variability across different topics. The findings indicated that while our approach is generally effective, it also encounters challenges when dealing with images that encompass overlapping thematic elements. Despite this, certain topics were clustered with high accuracy, showcasing the method's potential. The study also identified areas for improvement, particularly in refining the text generation and clustering algorithms to better handle the complexity of multiple themes within single images. In conclusion, this research contributes to the field of image data organization, providing insights into the use of textual descriptions for more contextually nuanced image clustering. It also opens avenues for future work to further improve the accuracy and interpretability of unsupervised learning methods in multimedia content analysis.
The research, "Reflection Removal and Facial Detection of Individuals in Vehicles," utilizes Single Image Reflection Removal (SIRR) technology and Face Detection to remove reflections and reduce glare caused by automotive glass and film. This enables the capture of facial images of individuals inside vehicles. SIRR technology enhances image quality by removing reflections from the surfaces of glass that might obscure objects. In this research, we explore the use of three models specialized in SIRR and YOLOv7 for Face Detection. However, the pre-trained models for reflection removal failed to effectively remove reflections and reduce glare from films. In this paper, we propose an approach to enhance the efficiency of removing reflections and reducing glare caused by automotive glass and film with opacities set at 40% and 60%, achieving an impressive improvement in Peak Signal-to-Noise Ratio (PSNR) by approximately 42.96% and Structural Similarity Index (SSIM) by approximately 34.16% compared to the pre-trained models.
This study presents the development and evaluation of a Web-Based Loan Management System incorporating smart contracts for a lending company. Focused on enhancing security, transparency, and efficiency in the lending process, the research integrates blockchain technology, Hyperledger Fabric, and API functionalities. Performance testing using GTMetrix ensures reliability. User acceptance testing, employing the User Experience Questionnaire (UEQ) and System Usability Scale (SUS), reveals positive pragmatic and hedonic qualities, with a final SUS score of 90, indicating excellent usability. The results suggest that the system, equipped with innovative technologies, holds promise for lending companies seeking secure and efficient financial solutions.
In view of the characteristics of near-space exploration data, such as complicated types, diverse sources, multiple disciplines and multiple parameters, a core metadata model of near-space exploration data is designed to meet the needs of data collection, data management and data sharing service. The mechanism of near-space exploration data management and sharing service is described in the lifecycle of near-space exploration data, including data collection and receiving, data classification, data quality control. The first domestic platform of near-space exploration data management and sharing service is developed based on the mechanism of near-space exploration data management and sharing service, and a high-performance spatio-temporal data sharing service framework integrating access control, query and retrieval, sharing and distribution, and statistical analyses is designed. The platform, which has collected 567 datasets with a total data volume of over 130 TB through practical application, provides an effective support for the follow-up exploration of near-space.
As technology advances, numerous businesses and institutions adopt the web application format. As the user base grows, concurrent visits increase, often leading to a subpar web application user experience. This study aims to investigate the performance factors affecting the MABIS web application system at various stages to optimize the web app's architecture, ensuring excellent accessibility, synchronization, scalability, and cost-efficiency. The optimization process for MABIS was guided by factors influencing the web system's performance, resulting in adjustments to align it more closely with system requirements. This research paper delves into the parameters for testing MABIS performance, the methods employed for performance testing, and strategies for performance enhancement. Building on this foundation, the study elaborates on optimization strategies from multiple perspectives, encompassing the backend, data storage, and gateway access, all within the context of a high-concurrency web application. The validated results demonstrate a notable improvement in system performance, increasing from 60% to 99%, while system structures improved by 5%, rising from 93% to 98%, ultimately earning an overall rating of "A" compared to the initial "C." This performance optimization process takes into account factors like load speed, response time, interactivity, responsiveness, and visual stability, leading to significantly enhanced results.
