Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi)
Not a member yet
1071 research outputs found
Sort by
Knowledge Management System Adoption Approach and the Critical Success Factors in Small Medium Enterprise: A Systematic Literature Review
Get PDFKnowledge is a substantial factor in an organization; therefore, the successful implementation of Knowledge Management (KM) or Knowledge Management System (KMS) is important for many organizations. This applies both for large companies and for companies categorized as Small Medium Enterprise (SME). How each company finds a solution to deal with KM problems, how to adopt KMS in its company structure, and what critical success factors (CSF) must be highlighted to implement those KMS often vary depending on the size of the organization. Regarding this issue, this study aims to find out how the adoption approach and CSF are used in the implementation of KM / KMS in SME. However, this study can also improve the state-of-the-art for KM / KMS implementation in SME and the important CSF in implementing it. In this review of the literature, a systematic review was performed with the steps as follows: (1) structure the research question, (2) define inclusion-exclusion criteria, (3) evaluation of paper quality, and (4) data extraction. The study found that in the last 5 years from the time when this research is conducted, TABLE 1 which is from 2016 to 2021, SME has been using many methods like training, meeting, sharing session, repository, and research as part of their KM / KMS adoption approach. We found also in the last 5 years that the CSF for implementing KM / KMS in SME is as follows: organization structure and flexibility, organization culture towards KM adoption, the quality of the knowledge, and communication within and across areas in the organization. communication within and across areas of the organization, and the team works within and across areas of the organization. SMEs can use this research as a guide to implement KM / KMS in their organization
Analysis and Mitigation of Religion Bias in Indonesian Natural Language Processing Datasets
No full textPrevious studies have shown the existence of misrepresentation regarding various religious identities in Indonesian media. Misrepresentations of other marginalized identities in natural language processing (NLP) datasets have been recorded to inflict harm against such marginalized identities in cases such as automated content moderation, and as such must be mitigated. In this paper, we analyze, for the first time, several Indonesian NLP datasets to see whether they contain unwanted bias and the effects of debiasing on them. We find that two of the three data sets analyzed in this study contain unwanted bias, whose effects trickle down to downstream performance in the form of allocation and representation harm. The results of debiasing at the dataset level, as a response to the biases previously discovered, are consistently positive for the respective dataset. However, depending on the data set and embedding used to train the model, they vary greatly at the downstream performance level. In particular, the same debiasing technique can decrease bias on a combination of datasets and embedding, yet increase bias on another, particularly in the case of representation harm.Previous studies have shown the existence of misrepresentation regarding various religious identities in Indonesian media. Misrepresentations of other marginalized identities in natural language processing (NLP) datasets have been recorded to inflict harm against such marginalized identities, in cases such as automated content moderation, and as such must be mitigated. In this paper, we analyze, for the first time, several Indonesian NLP datasets to see whether they contain unwanted bias and the effects of debiasing on them. We find that two, out of three, datasets analyzed in this study contain unwanted bias, whose effects trickle down to downstream performance under the form of allocation and representation harm. The results of debiasing at the dataset level, as a response to the biases previously discovered, are consistently positive for the respective dataset. Nevertheless, depending on the dataset and embedding used to train the model, they vary highly at the downstream performance level. In particular, the same debiasing technique can decrease bias on a combination of datasets and embedding, yet increase bias on another, particularly in the case of representation harm
Remote Penetration Testing with Telegram Bot
Get PDFThe widespread of websites and web applications makes them the main target of cyber attacks. One way to increase security is to perform a penetration test. This test is carried out using the attacker's point of view to find out vulnerabilities on a website or web application and then exploit these vulnerabilities. The results of the penetration test can be used as recommendations to close the gaps that have been known through testing. Because penetration testing requires special resources such as tools and operating systems, a solution is needed to make penetration testing possible with low resources. Telegram bots that are open source offer a solution to overcome these problems. Using the SDLC waterfall approach, this bot was built to provide penetration testing services by connecting the Kali Linux server as a tools provider and the Telegram bot as an interface to users. As a result, users can access penetration testing tools anywhere and anytime via the Telegram bot. To ensure that the bot can run well, testing is carried out through black box testing and load testing. Telegram bot is a solution for integrated compact automatic mobile penetration tester with low resources. Based on load testing, the maximum limit of users who can access Telegram bots simultaneously is 35 users with the highest load average of 5.4. Based on the results of the User Acceptance Test, the Telegram bot has an acceptance rate score of 88,457 % and a questionnaire score of 774 which is an agreed area.
The widespread of websites and web applications makes them the main target of cyber attacks. One way to increase security is to perform a penetration test. This test is carried out using the attacker's point of view to find out vulnerabilities on a website or web application and then exploit these vulnerabilities. The results of the penetration test can be used as recommendations to close the gaps that have been known through testing. Because penetration testing requires special resources such as tools and operating systems, a solution is needed to make penetration testing possible with low resources. Telegram bots that are open source offer a solution to overcome these problems. Using the SDLC waterfall approach, this bot was built to provide penetration testing services by connecting the Kali Linux server as a tools provider and the Telegram bot as an interface to users. As a result, users can access penetration testing tools anywhere and anytime via the Telegram bot. To ensure that the bot can run well, testing is carried out through black box testing and load testing. Telegram bot is a solution for integrated compact automatic mobile penetration tester with low resources. Based on load testing, the maximum limit of users who can access Telegram bots simultaneously is 35 users with the highest load average of 5.4. Based on the results of the User Acceptance Test, the Telegram bot has an acceptance rate score of 88,457 % and a questionnaire score of 774 which is an agreed area
Secure Cybersecurity Information Sharing for Sectoral Organizations Using Ethereum Blockchain and IPFS
Get PDFThe COVID-19 pandemic has resulted in increased cross-sector cyber-attacks. Passive and reactive cybersecurity techniques relying solely on technology are insufficient to combat sophisticated attacks, necessitating proactive and collaborative security measures to minimize attacks. Cybersecurity Information Sharing (CIS) enhances security via proactive and collaborative cybersecurity information exchange, but its implementation via cloud services faces threats from man in the middle (MITM) and distributed denial of service (DDoS) attacks, as well as a vulnerability in cloud storage involving centralized data control. These threats and vulnerabilities result in a lack of user confidence in the confidentiality, integrity, and availability of information. This paper proposes Secure Cybersecurity Information Sharing (SCIS) to secure Cybersecurity Information in sectoral organizations using the private interplanetary file system (IPFS) network and the private Ethereum Blockchain network. Private Ethereum Blockchain enables secure and transparent transaction logging, while Private IPFS network provides decentralized storage, addressing vulnerabilities in centralized storage systems. The outcomes of the tests reveal that the suggested SCIS system offers cybersecurity information availability, confidentiality, and integrity. SCIS provides a high level of security to protect cybersecurity information exchanged between sectoral organizations using the Private Ethereum Blockchain network and the Private IPFS network so that organizations can safely share and utilize information.
The COVID-19 pandemic has resulted in increased cross-sector cyber-attacks. Passive and reactive cybersecurity techniques relying solely on technology are insufficient to combat sophisticated attacks, necessitating proactive and collaborative security measures to minimize attacks. Cybersecurity Information Sharing (CIS) enhances security via proactive and collaborative cybersecurity information exchange, but its implementation via cloud services faces threats from man in the middle (MITM) and distributed denial of service (DDoS) attacks, as well as a vulnerability in cloud storage involving centralized data control. These threats and vulnerabilities result in a lack of user confidence in the confidentiality, integrity, and availability of information. This paper proposes Secure Cybersecurity Information Sharing (SCIS) to secure Cybersecurity Information in sectoral organizations using the private interplanetary file system (IPFS) network and the private Ethereum Blockchain network. Private Ethereum Blockchain enables secure and transparent transaction logging, while Private IPFS network provides decentralized storage, addressing vulnerabilities in centralized storage systems. The outcomes of the tests reveal that the suggested SCIS system offers cybersecurity information availability, confidentiality, and integrity. SCIS provides a high level of security to protect cybersecurity information exchanged between sectoral organizations using the Private Ethereum Blockchain network and the Private IPFS network so that organizations can safely share and utilize information
A Cost-Effective Vital Sign Monitoring System Harnessing Smartwatch for Home Care Patients
Get PDFPap smear is a digital image generated from the recording of cervical cancer cell preparation. Images generated are susceptible to errors due to relatively small cell sizes and overlapping cell nuclei. Therefore, an accurate analysis of the Pap smear image is essential to obtain the right information. This research compares nucleus segmentation and detection using gray-level cooccurrence matrix (GLCM) features in two methods: Otsu and polynomial. The data tested consisted of 400 images sourced from RepoMedUNM, a publicly accessible repository containing 2,346 images. Both methods were compared and evaluated to obtain the most accurate characteristics. The research results showed that the average distance of the Otsu method was 6.6457, which was superior to the polynomial method with a value of 6.6215. Distance refers to the distance between the nucleus detected by the Otsu and the Polynomial method. Distance is an important measure to assess how closely the detection results align with the actual nucleus positions. It indicates that the polynomial method produces nucleus detections that are on average closer to the actual nucleus positions compared to the Otsu method. Consequently, this research can serve as a reference for future studies in developing new methods to enhance identification accuracy
Analysis and Classification of Customer Churn Using Machine Learning Models
Get PDFAnalysis studies of customer loss (customer churn) have been used for years to increase profitability and build customer relationships with companies. Customer analysis using exploratory data analysis (EDA) to visualize data and the use of machine learning to classify customer churn are often used by past analysts. This study uses several machine learning models that can be used for customer churn classification, namely Logistic Regression, Random Forest, Support Vector Machine (SVM), Gradient Boosting, AdaBoost, and Extreme Gradient Boosting (XGBoost). However, there is a class imbalance factor in the dataset, which is the biggest challenge that analysts usually face in achieving good results in the classification of machine learning models. The Synthetic Minority Oversampling Technique (SMOTE) method is a popular method applied to deal with class imbalances in datasets. The results of the analysis show that the classification of churn customers using the XGBoost algorithm has the best level of accuracy compared to other algorithms, with an accuracy value of 0.829424, and the oversampling method with SMOTE tends to reduce the accuracy value of each classification algorithm. The Permutation Feature Importance (PFI) technique of the XGBoost model gets the result that tenure, monthly contracts, and TV streaming are the features that affect customer churn the most.Studi analisis terhadap kehilangan pelanggan (Customer Churn) telah digunakan selama bertahun-tahun untuk meningkatkan profitabilitas serta membangun hubungan pelanggan dengan perusahaan. Analisa pelanggan dan Penggunaan Machine Learning untuk klasifikasi terhadap Costumer Churn kerap digunakan oleh para analis terdahulu. Penelitian ini menggunakan beberapa model Machine Learning yang dapat digunakan untuk klasifikasi Costumer Churn yaitu Logistic Regression, Random Forest, Support Vector Machine (SVM), Gradient Boosting, AdaBoost, dan Extreme Gradient Boosting (XGBoost). Terdapat faktor ketidakseimbangan kelas pada dataset yang merupakan tantangan terbesar biasa dihadapi para analis untuk mendapatkan hasil klasifikasi model Machine Learning yang baik. Metode Synthetic Minority Over-sampling Technique (SMOTE) merupakan metode yang populer diterapkan dalam rangka menangani ketidakseimbangan kelas pada dataset. Dari hasil analisa didapatkan bahwa klasifikasi pelanggan churn menggunakan algoritma XGBoost memiliki tingkat akurasi terbaik dibandingkan dengan algoritma lainnya dengan nilai akurasi mencapai 0.829424, dan metode oversampling dengan SMOTE cenderung menurunkan nilai akurasi pada tiap algoritma klasifikasi
Pengenalan Tulisan Tangan Aksara Bali Menggunakan Faster R-CNN
No full textIn Balinese culture, the ability to read Balinese script is one of the challenges young generations face. Advances in machine learning have proposed handwriting detection systems using both traditional and deep learning models. However, the traditional approach is usually impractical and is prone to inaccurate identification results. Convolutional neural network (CNN)-based models integrate feature extraction and classification into an end-to-end pipeline to increase performance. This research proposes that recognizing characters through an object detection approach makes an end-to-end process of localizing and classifying several characters simultaneously using the Faster R-CNN. Four CNN models, including ResNet-50, ResNet-101, ResNet-152, and Inception ResNet V2, were tested to detect 28 Balinese characters in a single form that covers 18 consonants and 10 digits using Intersection over Union (IoU) thresholds: 0.5 and 0.75. ResNet-50 and Inception ResNet V2 achieve 0.991 mAP at IoU of 0.5, while Inception ResNet V2 excels at IoU of 0.75. Further analysis showed that the class ‘nol’ had the lowest Recall due to many undetected ground truths. Meanwhile, class ‘ba’ had the lowest Precision due to its similarity to classes “ga” and “nga”. This research contributes to the experiment with Faster R-CNN in detecting handwritten Balinese scripts.Dalam budaya Bali, kemampuan membaca aksara Bali menjadi salah satu tantangan yang dihadapi generasi muda. Upaya pengenalan tulisan tangan aksara Bali telah dilakukan dengan menggunakan model pembelajaran tradisional dan mendalam. Namun, model tradisional memerlukan beberapa tahapan: preprocessing, segmentasi, ekstraksi fitur, dan klasifikasi. Sementara itu, model berbasis Convolutional Neural Network (CNN) mengintegrasikan ekstraksi fitur dan klasifikasi ke dalam pipeline end-to-end. Penelitian ini mengusulkan metode untuk menambahkan segmentasi ke dalam pipa end-to-end dengan menerapkan model Faster R-CNN. Empat model CNN (ResNet-50, ResNet-101, ResNet-152, dan Inception ResNet V2) diuji untuk mendeteksi 28 karakter bahasa Bali dalam bentuk tunggal yang meliputi 18 konsonan dan 10 digit menggunakan dua titik temu di atas Union (IoU) ambang batas 0,5 dan 0,75. Hasil menunjukkan bahwa ResNet-50 dan Inception ResNet V2 mencapai nilai mAP 0,991 pada ambang batas IoU 0,5. Namun, Inception ResNet V2 unggul di ambang batas IoU yang lebih tinggi. Analisis lebih lanjut menunjukkan bahwa kelas "nol" memiliki nilai ingatan terendah karena banyak kebenaran dasar yang tidak terdeteksi. Sebaliknya, kelas “ba” memiliki nilai presisi paling rendah karena kemiripannya dengan kelas “ga” dan “nga”. Penelitian ini berkontribusi untuk bereksperimen dengan model deteksi objek dalam mendeteksi aksara Bali tulisan tangan
Imputation Missing Value to Overcome Sparsity Problems in The Recommendation System
No full textA recommendation system is a system that provides suggestions or recommendations to a product or service for its users. One of the problems encountered in the recommendation system is sparsity, namely the lack of available data for analysis, resulting in poor performance of the recommendation system because it cannot provide the proper recommendations. On this basis, this study proposes the mean method and the stochastic hot-deck method to calculate missing values to improve the quality of the recommendations. The experimental results show that the hot-deck imputation method gives better results than the mean imputation method with smaller RMSE and MAE values, namely 2,706 and 2,691.Sistem rekomendasi adalah suatu sistem yang memberikan saran atau rekomendasi terhadap suatu produk atau layanan untuk para penggunanya. Salah satu tantangan dalam pengembangan sistem rekomendasi adalah masalah sparsity, yaitu kurangnya data yang tersedia untuk melakukan analisis. Sparsity dapat mengakibatkan performa sistem rekomendasi menjadi buruk karena tidak dapat memberikan rekomendasi yang tepat. Salah satu metode yang dapat digunakan untuk menyelesaikan masalah sparsity adalah imputasi. Imputasi adalah suatu proses pengisian data kosong atau missing value pada dataset dengan memperkirakan nilai yang hilang berdasarkan data yang tersedia. Penelitian menggunakan dua metode imputasi missing value, yaitu metode mean dan metode stochastic hot-deck. Hasil eksperimen menunjukkan metode imputasi hotdeck memberikan hasil yang lebih baik daripada metode imputasi mean pada dataset yang digunakan, dengan nilai RMSE dan MAE yang lebih kecil yaitu 2.706 dan 2.691
Artificial Neural Network-Based Prediction Model Back Propagation on Blood Demand and Blood Supply
Get PDFThe balance between blood demand and supply at the Indonesian Red Cross Blood Transfusion Unit (UTD-PMI) is crucial. This condition must be maintained to reduce unused or expired blood supplies. Despite the situation in UTD-PMI, where the blood supply exceeds demand, there is still a shortage of blood when needed by patients. This research aims to model the prediction of blood demand and supply for each blood type using the Back Propagation artificial neural network approach. Data from the last 3 years, from 2020 to 2022, were utilized in this research process. There are three stages in this research process. The first stage involves the training process, using data from January 2020 to December 2021. The testing process utilizes data from January 2021 to December 2022. The prediction process involves displaying the forecasted data for the next 12 months from January to December 2023. The accuracy of the calculations is assessed using the mean square error (MSE). Ultimately, the research results present the prediction model for the four types of blood with respect to the demand and supply. These findings can serve as a reference to regulate future blood donation activities carried out by the UTD-PMI
Forensic Analysis of Faces on Low-Quality Images using Detection and Recognition Methods
Get PDFFacial recognition is an essential aspect of conducting criminal action investigations. Captured images from the camera or the recording video can reveal the perpetrator's identity if their faces are deliberately or accidentally captured. However, many of these digital imagery results display the results of image quality that is not good when seen by the human eye. Hence, the facial recognition process becomes more complex and takes longer. This research aims to analyze face recognition on a low-quality image with noise, blur and brightness problem to help digital forensic investigator do an investigation in recognizing faces that the human eye can’t do. The Viola-Jones algorithm method has several processes such as the Haar feature, integral image, adaboost, and cascade classifier for detecting a face in an image. Detected face will be passed to the next process for recognition call Fisher’s Linear Discriminant (FLD), Local Binary Pattern’s (LBP) and Principal Component analysis (PCA). The software's facial recognition feature shows one of the images in the database that the program suspects has the same face as the analyzed face image. In conclusion, from the analysis we determined that LBP approach is the best among the other recognition methods for blur and brightness problem, bet found PCA method is the best for recognize face in noise problem. The software's facial recognition feature shows one of the images in the database that the program suspects has the same face as the analyzed face image. The position of the face object in the image, whether or not there is an additional object that was not previously included in the image in the dataset, as well as the brightness level of an image and the color of the face's skin, all affect the accuracy rates.Facial recognition is an essential aspect of conducting criminal action investigations. Captured images from the camera or the recording video can reveal the perpetrator's identity if their faces are deliberately or accidentally captured. However, many of these digital imagery results display the results of image quality that is not good when seen by the human eye. Hence, the facial recognition process becomes more complex and takes longer. This research aims to analyze face recognition on a low-quality image with noise, blur and brightness problem to help digital forensic investigator do an investigation in recognizing faces that the human eye can’t do. The Viola-Jones algorithm method has several processes such as the Haar feature, integral image, adaboost, and cascade classifier for detecting a face in an image. Detected face will be passed to the next process for recognition call Fisher’s Linear Discriminant (FLD), Local Binary Pattern’s (LBP) and Principal Component analysis (PCA). The software's facial recognition feature shows one of the images in the database that the program suspects has the same face as the analyzed face image. In conclusion, from the analysis we determined that LBP approach is the best among the other recognition methods for blur and brightness problem, bet found PCA method is the best for recognize face in noise problem. The software's facial recognition feature shows one of the images in the database that the program suspects has the same face as the analyzed face image. The position of the face object in the image, whether or not there is an additional object that was not previously included in the image in the dataset, as well as the brightness level of an image and the color of the face's skin, all affect the accuracy rates