International Journal on Information and Communication Technology (IJoICT)

UI/UX Design for Student Discussion Applications Based Felder Silverman Learning Style with the Design Thinking Method

HNW Syahuda Nahatmasuni, Anisa Herdiani, Ati Suci Dian Martha — Tue, 12 Dec 2023 09:56:31 +0700

Telkom University is one of the educational institutions that implement asynchronous learning through its Learning Management System (LMS). Based on preliminary research conducted with 37 respondents, just 13.5% of the participants LMS using a smartphone, because responsiveness issue. The Discussion Forum is a frequently used feature with high student interaction. However, this feature has several shortcomings, such as the lack of responsiveness in the mobile web interface and the limited interaction between users and professors. This research will employ the Design Thinking methodology and adopt the Felder Silverman Learning Style (FSLS). The evaluation of the prototype design resulted in a System Usability Scale (SUS) testing score of 72.22 for the LMS Celoe website and 85.65 for the proposed UI/UX application. The SUS testing score for the LMS Celoe website falls within Quadrant C, indicating an acceptable level of acceptance with a grade C scale and a rating of "Good." On the other hand, the SUS testing score for the proposed UI/UX application falls within Quadrant B, with an acceptable level of acceptance, a grade B scale, and an "Excellent" rating.

The Development of High Availability Database Infrastructure for OSS Projects with Monitoring Systems in Cloud Computing Environments

Akmal Ikhsan, Dana Kusumo — Tue, 12 Dec 2023 13:02:04 +0700

This study addressed a critical problem in the Kominfo's Operation Support System (OSS) project, which significantly impacted business operations. The outage of the MongoDB database, a vital component of OSS, resulted in server crashes and data damage. To overcome this issue, the OSS team initiated a project to migrate to PostgreSQL for high availability and improved database performance. The implementation involved the use of HA PostgreSQL technology, with multiple connected servers sharing data in real-time. Through functional and performance testing, the study has demonstrated that the HA PostgreSQL system increased database availability, managed server failures, and facilitated effective cluster administration. The findings of this research can guide the development of the OSS project's IT infrastructure and serve as a reference for similar projects utilizing HA PostgreSQL technology.

Analysis The Impact of E-Service Quality on E-Customer Satisfaction in Cinema Ticket Booking Application

Ditya Ilmi Rizqi, Rio Guntur Utomo, Muhammad Al Makky — Mon, 11 Dec 2023 10:34:57 +0700

The development of technology and information has made everything easier and more efficient for us to obtain. One of its applications is online ticket booking in cinema networking applications. Innovations in the implementation of this technology can attract a significant number of buyers as it is considered to facilitate users in ticket booking transactions. Cinema networking service providers offer various features and information on their applications to attract buyers and create a positive image and trust for users to reuse their online cinema service applications. In order to maintain and enhance user satisfaction, service providers must also improve the quality of the services provided. This research aims to examine the influence of service quality on user satisfaction. Data collection was conducted through questionnaires distributed to respondents who are users of cinema ticket booking applications. The data processing technique used is SmartPLS (Smart Partial Least Squares) to analyze the measurement and structural models. The method employed is E-Service Quality with seven dimensions as indicators, namely Efficiency, Fulfillment, Reliability, Privacy, Responsiveness, Compensation, and Contact. The results of this study indicate the influence of E-Service Quality variables on E-Customer Satisfaction variables.

Sentiment Analysis on Acute Kidney Syrup Videos Using CNN and LSTM Algorithms

Guido Tamara, Kemas Muslim L — Tue, 12 Dec 2023 13:31:33 +0700

The issue of acute kidney failure, particularly caused by the consumption of cough syrup, was circulating around October 2022 and has become a serious public health concern. This issue has drawn extensive attention and sparked various reactions on social media. In this digital era, public opinion expressed in comments on social media platforms like YouTube significantly impacts societal perceptions. Therefore, in the context of the aforementioned issue, sentiment analysis on YouTube video comments can provide valuable insights into societal perceptions and people’s reactions. Therefore, this study focuses on the sentiment analysis of public opinions expressed in YouTube comments related to this matter. The methods employed for this analysis include Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) with Word2Vec feature extraction. The findings of this study indicate that both these methods produce good performance results with an oversampling dataset with a 90:10 data proportion. In the performance comparison, CNN yielded the highest accuracy, at 0.92, while LSTM was at 0.90.

Machine Learning Sentiment Analysis in Cyber Threat Intelligence Recommendation System

Marastika wicaksono aji bawono aji bawono, Sachlany Kasman , Stevani Dwi Utomo — Thu, 14 Dec 2023 15:03:26 +0700

The use of the digital world is increasing every day. Attacks and data theft occur on various websites, both government-owned and commercial and banking sites. Therefore, this research aims to identify the threats of frequently occurring viruses in a country. There is a considerable amount of news explaining cybercrime incidents. The problem of this research is that unstructured data such as articles and technical reports are difficult to analyze and identify the types of cybercrime attacks. Previous research attempted to semantically extract unstructured cyber threats, but there were shortcomings in previous research. The novelty of this research is the development of a Cyber Threat Intelligence (CTI) machine learning model to identify the types of virus attacks or cybercrimes that frequently occur in e-commerce transactions, so that they can take rescue actions for incident handling in the digital world using tactics, techniques, and procedures (TTP). The method involves using machine learning, taking Cyber Threat Intelligence (CTI) documents as input regarding cybersecurity threat handling steps, and then processing the data using AI TF-IDF and Bags of Words for the identification of steps, tactics, techniques, and procedures required for each frequently occurring security incident.

Socio-user Context Aware-Based Recommender System: Context Suggestions for A Better Tourism Recommendation

Kusuma Adi Achmad, Lukito Edi Nugroho, Achmad Djunaedi, Widyawan — Mon, 25 Dec 2023 16:12:22 +0700

The existing tourism recommender system model is mostly predictive analytics for destination recommendations (item recommendation). Limited research has been conducted in the discussion of a recommender system model, particularly context suggestion. Thus, it is necessary to develop a recommender system model not only to predict tourism destinations but also to suggest contexts appropriate for tourist preferences (context suggestions). A deep learning method was used to create a model of the socio-user context aware-based recommender system for context suggestions. The attribute used as a label to suggest context was uHijos, uCuisine, uAmbience, and uTransport. The accuracy of the socio-user context aware-based recommender system in suggesting the context of uHijos, uAmbience, and uTransport was 100% with an error rate of 0%. It was found that only the level of recognition of the model in suggesting uCuisine was less accurate (below 30%) with a classification error for more than 70%. Performance evaluation of the socio-user model context-based recommender system was considered efficient, particularly for the evaluation of the level of accuracy, completeness (recall/sensitivity), precision, and a harmonic average of precision and recall (F-score), mainly for label/context of uHijos, uAmbience, and uTransport.

XGBoost for Predicting Airline Customer Satisfaction Based on Computational Efficient Questionnaire

Nur Ghaniaviyanto Ramadhan, Aji Gautama Putrada — Fri, 29 Dec 2023 11:02:01 +0700

Customer satisfaction can be created through a well-crafted service quality strategy, which forms the cornerstone of a successful business-customer relationship. Establishing and nurturing these relationships with customers is vital for long-term success. Within the airline industry, a persistent challenge lies in enhancing the passenger experience during flights, necessitating a comprehensive understanding of customer demands. Addressing this challenge is crucial for airlines aspiring to thrive in a competitive landscape, thus underlining the significance of providing top-notch services. This study addresses this issue by leveraging predictive airline customer satisfaction data analysis. We forecast customer satisfaction levels using a powerful Extreme Gradient Boosting (XGBoost) ensemble-based model. An integral aspect of our methodology involves handling missing values in the dataset, for which we utilize mean-value imputation. Furthermore, we introduce a novel logistic Pearson Gini (Log-PG) score to identify the factors that significantly influence airline customer satisfaction. In our predictive model, we achieved notable results, showing an accuracy and precision of 0.96. To ascertain the efficiency of our model, we conducted a comparative analysis with other boosting-type ensemble prediction models, such as gradient boosting and adaptive boosting (AdaBoost). The comparative assessment established the superiority of the XGBoost model in predicting airline customer satisfaction.

Predicting Forest Fire Hotspots with Carbon Emission Insights Using Random Forest and Gradient Boosting Regression

irma palupi, bambang ari wahyudi, Naila AL Mamuda, Ayu Shabrina — Fri, 29 Dec 2023 14:22:07 +0700

This research paper focuses on predicting the dispersion of carbon emissions, a crucial indicator for identifying potential forest fire hotspots in the wooded regions of Sumatra Island, Indonesia. Forest fires, often triggered by extended periods of dry weather, result in significant environmental degradation, impacting both the ecosystem and the economy. Furthermore, health concerns arise from smoke inhalation, leading to respiratory problems. To achieve this predictive capability, we harnessed valuable datasets, including GFED4.1s for carbon emissions and ERA5 for historical climate indicators, spanning from 1998 to 2022. Employing supervised learning ensemble methods, specifically Random Forest Regression (RFR) and Gradient Boosting Regression (GBR), we sought to forecast carbon emissions. It is noteworthy that our predictions encompassed carbon emission values from 1998 to 2023, providing insights into recent trends. Our analysis showed that GBR did better than RFR in terms of evaluation metrics, with a root mean square error (RMSE) of 10.87 and a mean absolute error (MAE) of 2.91. This was done by carefully tuning the hyperparameters. Additionally, our study highlighted that precipitation, temperature, and humidity were the primary climate factors influencing carbon emission values.

Reducing Lending Risk: SVM Model Development with SMOTE for Unbalanced Credit Data

Josya Ryan Alexandro Purba, Qilbaaini Effendi Muftikhali, Bony Parulian Josaphat — Sat, 30 Dec 2023 11:22:23 +0700

Lending is an important activity for banks in managing available funds. However, lending is also an activity that has a high risk, because not all customers who borrow funds can fulfill the responsibilities of the existing agreement. Because of this, it is necessary to have a method that can predict creditworthiness to customers in order to minimize the risks that arise. This research uses machine learning method, namely Support Vector Machine (SVM) in predicting creditworthiness. This method is applied and compared before and after the Synthetic Minority Oversampling Technique (SMOTE) on historical bank credit data BPR NBP 16 Rantau Prapat, North Sumatra and find the best parameters with grid search. According to the results of the analysis based on Area Under the Receiver Operating Characteristic Curve (AUC-ROC), SVM with SMOTE shows better results, namely 96%, than SVM without SMOTE, namely 56%.

Performance of Time-Based Feature Expansion in Developing ANN Classification Prediction Models on Time Series Data

Sri Suryani Prasetiyowati , Arnasli Yahya, Aniq Atiqi Rohmawati — Sun, 31 Dec 2023 15:13:42 +0700

The prediction problem in most research is the main goal, to estimate future events related to the field under study. Research on classification that involves the prediction process in it, with spatial-time data and influenced by many features, such as the problem of disease spread, climate change, regional planning, environment, economic growth, requires methods that can predict while solving the problem of features and time. To obtain a time-based classification prediction model using many features, this research uses machine learning methods, one of which is Artificial Neural Network (ANN). The scenario carried out is to develop a t+r classification prediction model by expanding features based on the time t-r of the previous period. The performance of feature expansion in the development of ANN classification prediction models is determined based on the optimal accuracy value of the combination of t-r classification prediction models for the previous time period. By implementing the model on the data, it is found that the performance of time-based feature expansion in ANN classification ranges from 3.5% to 11%. While the optimal accuracy value is obtained from the feature expansion scenario of 3 to 5 time periods earlier.

Hoax Detection of Covid-19 News on Social Media using Convolutional Neural Network (CNN) and Support Vector Machine (SVM)

Arvia Dwi Cahyani, Andi Kholik Ramdani — Sun, 31 Dec 2023 16:39:14 +0700

It is undeniable that nowadays news spreads very quickly on social media. The ease of getting news on social media has resulted in some users using and spreading news without knowing the authenticity of the news. Twitter (X) users play an important role in spreading news on social media. In early 2020, cases of Covid-19 started to occur in Indonesia and some people spread news about Covid-19 without knowing the real information. The news is increasingly spreading through Twitter media which is shared by irresponsible people. This research builds a system that can detect hoax news on social media. The stages in this study started from crawling data, data preprocessing, word embedding, data separation, modeling process, and model evaluation. The methods used are Convolutional Neural Network (CNN) and Support Vector Machine (SVM). The dataset used is news of Covid-19 in X Social media. The experiment showt that the use of the N-Gram Unigram + Bigram + Trigram combination on CNN produces an accuracy value of 75.8%, meanwhile in the SVM modeling produces 77.9%. It can be concluded that SVM has better performance than CNN in detecting hoax news,

Movie Recommendation System Based on Synopsis Using Content-Based Filtering with TF-IDF and Cosine Similarity

Armadhani Hiro Juni Permana Juni Permana, Agung Toto Wibowo — Fri, 08 Dec 2023 14:59:44 +0700

Recommendation systems have become an interesting topic in the field of artificial intelligence and data analysis. In the current era of technological advancement, the entertainment industry is rapidly growing, particularly the film industry, which is highly popular among the public due to their enthusiasm for watching movies. The increasing number and variety of films with various genres and titles have made it challenging for users to choose a film. To assist them in selecting movies, the presence of a recommendation system is necessary to provide information or film recommendations based on user interests and preferences. In this research, the development of the recommendation system will utilize the content-based filtering method, employing the TF-IDF algorithm and cosine similarity. The dataset used in this study is derived from publicly available data (MovieLens). The results of this research demonstrate that the TF-IDF and cosine similarity algorithms provide recommendations that align with the viewers' interests, as measured by precision, recall, and f1-score calculations.

Convolutional Neural Network Implementation with AlexNet Architecture for Face Recognition

Denny Hardiyanto, Dyah Anggun Sartika, Imam Junaedi, Sukamto — Thu, 14 Dec 2023 11:32:42 +0700

In today's digital era, the process of facial recognition has a very big role. Face recognition has benefits for authentication and identification processes. The development of facial recognition research continues to be carried out with the aim of being able to get the right algorithm, more accurate, faster processing, to be able to recognize faces from various angles. In this study, a performance test was performed on the Convolutional Neural Network (CNN) algorithm with the AlexNet architecture, which is one of the deep learning algorithm developments for facial recognition. AlexNet has 8 convolution layers so that it will not leave even the slightest feature of the object. The process of training and testing the system uses the MATLAB programming language. The number of datasets used is 400 image data which is divided into 360 training image data and 40 test image data. The 400 data come from 4 classes of facial images that have been labeled with names and each classes have 100 images. The training process produces an accuracy of 100% and the testing process produces an accuracy of 95%.

The The Recognition of American Sign Language Using CNN with Hand Keypoint

Muhamad Asep Ridwan, Aradea, Husni Mubarok — Sat, 16 Dec 2023 20:10:26 +0700

Sign Language is a method used by the deaf community for their communication. In line with the advances of deep learning, researchers have widely interpreted neural networks for language recognition in recent years. Many models and hardware have been developed to help get high accuracy in language recognition, but generally, the problem of accuracy is still a concern of researchers, even the accuracy problem related to American language or American sign language (ASL) still requires further research to solve. This paper discusses a method to improve ASL recognition accuracy using Convolutional Neural Network (CNN) with hand keypoint. Pre-trained Keypoint detector is used to generate hand keypoints on the massey dataset as an input for classification in the CNN model. The results show that the accuracy of the proposed method is better than the previous studies, obtaining an accuracy of 99.1% in recognizing the 26 statistical signs of the ASL alphabet.

Performance Analysis of Facial Image Feature Extraction Algorithm for Smart Home Security System Detection

Muhammad Ihsan Adly, Satria Mandala — Mon, 08 Jan 2024 10:54:09 +0700

Alongside the development of technology to facilitate multi-family security, security tools are also being developed. Smart home security is one of the very popular security tools in Indonesian home construction. The tool works automatically in real time and has no restrictions on environmental conditions. However, currently available tools still lack consistent accuracy and consistent performance. To solve this problem, the author proposes a smart home security system with an Arduino UNO-connected camera, two relay modules, a magnetic lock, and connecting to a home Internet of Things system. The methods used in the research for this thesis project were: 1. Literature review of ongoing Smart Home Security using facial image feature extraction algorithm research; 2. Deployment of Arduino UNO, 2 Relay Module, and Solenoid Lock; 3. The feature extraction algorithm used is Wavelet. The proposed method is expected to achieve an accuracy of 80% or more. The experimental results showed that the proposed prototype of this experiment achieved the accuracy of 85.7%. In addition to accuracy, there is also precision rate at 87.94%, recall rate at 87.56%, and f1-score rate at 87.28%