Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi)
Not a member yet
1071 research outputs found
Sort by
Ekstraksi Fitur untuk Peningkatan Klasifikasi Teks Komentar Video Youtube Spam Menggunakan Deep Learning
No full textThe proposed algorithms are Bidirectional Long Short Term Memory (BiLSTM) and Conditional Random Fields (CRF) with Data Augmentation Technique (DAT). DAT integrates spam YouTube video comments into the traditional TF-IDF algorithm and generates a weighted word vector. The weighted word vector is fed into BiLSTM CRF to capture context information effectively. The result of this study is a new classification model to spam YouTube comment videos and increase the computational value of its performance. This research conducted two experiments: the first using BiLSTM CRF without DAT and the second using BiLSTM CRF with DAT. The experimental results state that the evaluation score using BiLSTM CRF with DAT shows outstanding performance in text classification, especially in spam YouTube video comment texts, with accuracy = 83.3%, precision = 83.6%, recall = 83.3%, and F-measure = 83.3%. So the combination of the BiLSTM-CRF method and the Data Augmentation Technique is very precise, so it can be used to increase the accuracy of classification texts for spam YouTube video commentsAlgoritma yang diusulkan adalah Bidirectional Long Short Term Memory (BiLSTM) dan Conditional Random Fields (CRF) dengan Data Augmentation Technique (DAT). DAT mengintegrasikan komentar video youtube spam ke dalam algoritme TF-IDF tradisional dan menghasilkan vektor kata berbobot. Vektor kata yang diberi bobot dimasukkan ke BiLSTM CRF untuk menangkap informasi konteks secara efektif. Hasil dari penelitian ini adalah model klasifikasi baru untuk spam video komentar youtube dan meningkatkan nilai komputasi kinerjanya. Penelitian ini melakukan dua percobaan, percobaan pertama menggunakan BiLSTM CRF tanpa DAT, percobaan kedua menggunakan BiLSTM CRF dengan DAT. Hasil percobaan menyatakan bahwa skor evaluasi menggunakan BiLSTM CRF dengan DAT menunjukkan kinerja yang luar biasa dalam klasifikasi teks, khususnya pada teks komentar video spam youtube dengan akurasi = 83,3% presisi = 83,6%; recall = 83,3% dan F-Measure = 83,3%. Sehingga kombinasi metode BiLSTM-CRF dan Data Augmentation Technique sangat tepat, sehingga dapat digunakan untuk meningkatkan akurasi klasifikasi teks komentar video youtube spa
Tajweed-YOLO: Object Detection Method for Tajweed by Applying HSV Color Model Augmentation on Mushaf Images
Get PDFTajweed is a basic knowledge of learning to read the Al-Qur’an correctly. Tajweed has many laws grouped into several parts so that only some people can memorize and implement Tajweed properly. Therefore, it is necessary to have an automatic detection system to facilitate the recognition of Tajweed, which can be used daily. This study presents Tajweed-YOLO, which applies the HSV color augmentation model to detect Tajweed objects in Mushaf images using YOLO. The contribution to this study was to compare the three versions of You Only Look Once (YOLO), i.e., YOLOv5, YOLOv6, and YOLOv7, and usage of the HSV color model augmentation to improve Tajweed detection performance. Comparing the three YOLO versions aims to solve problems in detecting small objects and recognizing various forms of Mushaf writing fonts in Tajweed detection. Meanwhile, the HSV color model aims to recognize Tajweed objects in various Mushaf and handle minority class problems. In this study, we collected four different Al-Qur’an mushaf with 10 Tajweed classes. The augmentation process can increase the detection performance by up to 85% compared to without augmentation 6th Class (Mad Jaiz Munfashil) using YOLOv6. The comparison of three YOLO versions concluded that YOLOv7 was better than YOLOv5 and YOLOv6, seen in data with augmentation and without augmentation. The evaluation results of mAP0.5 on 17 test data on the YOLOv7, YOLOv6, and YOLOv5 models are 80%, 69%, and 71%, respectively. These results prove that this research model’s results are suitable for the real-time detection of Tajweed.
Tajweed is a basic knowledge of learning to read the Al-Qur’an correctly. Tajweed has many laws grouped into several parts so that only some people can memorize and implement Tajweed properly. Therefore, it is necessary to have an automatic detection system to facilitate the recognition of Tajweed, which can be used daily. This study presents Tajweed-YOLO, which applies the HSV color augmentation model to detect Tajweed objects in Mushaf images using YOLO. The contribution to this study was to compare the three versions of You Only Look Once (YOLO), i.e., YOLOv5, YOLOv6, and YOLOv7, and usage of the HSV color model augmentation to improve Tajweed detection performance. Comparing the three YOLO versions aims to solve problems in detecting small objects and recognizing various forms of Mushaf writing fonts in Tajweed detection. Meanwhile, the HSV color model aims to recognize Tajweed objects in various Mushaf and handle minority class problems. In this study, we collected four different Al-Qur’an mushaf with 10 Tajweed classes. The augmentation process can increase the detection performance by up to 85% compared to without augmentation 6th Class (Mad Jaiz Munfashil) using YOLOv6. The comparison of three YOLO versions concluded that YOLOv7 was better than YOLOv5 and YOLOv6, seen in data with augmentation and without augmentation. The evaluation results of mAP0.5 on 17 test data on the YOLOv7, YOLOv6, and YOLOv5 models are 80%, 69%, and 71%, respectively. These results prove that this research model’s results are suitable for the real-time detection of Tajweed
Detecting Diseases on Clove Leaves Using GLCM and Clustering K-Means
No full textThe detection of disease in clove plant leaves is generally carried out by diagnosing the symptoms that appear on clove plants. This diagnosis is conducted by clove farmers only by relying on their experience or even having to seek information from other clove farmers. This is because the agricultural sector has no disease detection system for clove leaves by utilizing digital image processing technology to detect diseases in clove leaves. In this study, the researchers applied two methods to make it easier for clove farmers to diagnose diseases in their clove plants. Those methods were the imaging system using Gray Level Co-Occurrence Matrix (GLCM) and disease clustering using the K-Means algorithm. The objective of this study was to design and build image pattern recognition by utilizing 4 features of the GLCM: energy, entropy, homogeneity, and contrast. These 4 features were used to obtain the extraction value from an image. The outcomes were then used to cluster the clove plant diseases using the K-Means method. In making the software, the researchers used Javascript, HTML, CSS, PHP, and MySql to create a database. The output in this study was an information system application that provides disease-type clustering using the K-Means algorithm. The results of the GLCM concerning extracting images of clove plant leaves affected by disease indicated that the created system can be used to help clove farmers in diagnosing what diseases are infecting their plants by only uploading photos from affected leaves of the clove plant. Furthermore, the results of the K-Means calculation on the examined data showed several categories of Anthracnose leaf spot diseases. In addition, sample number #40 was included in cluster 2 status, in which the average values for energy, entropy, homogeneity, and contrast were 0.583, 0.175, 0.939, and 0.175, respectively.The detection of disease in clove plant leaves is generally carried out by diagnosing the symptoms that appear on clove plants. This diagnosis is conducted by clove farmers only by relying on their experience or even having to seek information from other clove farmers. This is because the agricultural sector has no disease detection system for clove leaves by utilizing digital image processing technology to detect diseases in clove leaves. In this study, the researchers applied two methods to make it easier for clove farmers to diagnose diseases in their clove plants. Those methods were the imaging system using Gray Level Co-Occurrence Matrix (GLCM) and disease clustering using the K-Means algorithm. The objective of this study was to design and build image pattern recognition by utilizing 4 features of the GLCM: energy, entropy, homogeneity, and contrast. These 4 features were used to obtain the extraction value from an image. The outcomes were then used to cluster the clove plant diseases using the K-Means method. In making the software, the researchers used Javascript, HTML, CSS, PHP, and MySql to create a database. The output in this study was an information system application that provides disease-type clustering using the K-Means algorithm. The results of the GLCM concerning extracting images of clove plant leaves affected by disease indicated that the created system can be used to help clove farmers in diagnosing what diseases are infecting their plants by only uploading photos from affected leaves of the clove plant. Furthermore, the results of the K-Means calculation on the examined data showed several categories of Anthracnose leaf spot diseases. In addition, sample number #40 was included in cluster 2 status, in which the average values for energy, entropy, homogeneity, and contrast were 0.583, 0.175, 0.939, and 0.175, respectively
Factors Affecting the PeduliLindungi User Experience Based on UX Honeycomb
Get PDFWith background of Covid-19 pandemic, Indonesian state trying to make various efforts, so people comply health protocols. One of them is through PeduliLindungi application. PeduliLindungi has 3 main functions, namely tracing, tracking, warning and fencing. However, PeduliLindungi is deemed unable to meet user needs in terms of appearance and experience provided. This study aims to find factors that affect user experience in PeduliLindungi app based on UX Honeycomb. UX Honeycomb is a tool that can explain various aspects of user experience design in 7 indicators and grouped into 3 variables. The 3 variables are Think (useful, valuable, credible), Feel (desirable, credible), and Use (findable, accessible, usable). This study uses primary data by distributing online questionnaires to 404 respondents contains 15 statements that represent all UX Honeycomb variables, with 5 scales of answer choices namely strongly disagree/disagree/neutral/agree/strongly agree. From the calculation results, it is found that all variables and indicators significantly affect user experience with greatest level of influence being on the Think variable at 0,418, the second Use at 0,219, and the last is Feel at 0,151. Further research is expected can measure the level of influence on user experience by making direct comparisons with similar health apps.
With background of Covid-19 pandemic, Indonesian state trying to make various efforts, so people comply health protocols. One of them is through PeduliLindungi application. PeduliLindungi has 3 main functions, namely tracing, tracking, warning and fencing. However, PeduliLindungi is deemed unable to meet user needs in terms of appearance and experience provided. This study aims to find factors that affect user experience in PeduliLindungi app based on UX Honeycomb. UX Honeycomb is a tool that can explain various aspects of user experience design in 7 indicators and grouped into 3 variables. The 3 variables are Think (useful, valuable, credible), Feel (desirable, credible), and Use (findable, accessible, usable). This study uses primary data by distributing online questionnaires to 404 respondents contains 15 statements that represent all UX Honeycomb variables, with 5 scales of answer choices namely strongly disagree/disagree/neutral/agree/strongly agree. From the calculation results, it is found that all variables and indicators significantly affect user experience with greatest level of influence being on the Think variable at 0,418, the second Use at 0,219, and the last is Feel at 0,151. Further research is expected can measure the level of influence on user experience by making direct comparisons with similar health apps
Smoke Automation and Regression Testing on a peer-to-peer lending Website with the Data-DrivenTesting Method
Get PDFSoftware testing is considered as one of the most important processes in software development as it checks whether the system meets the requirements and specifications of the users. The testing process manually or automatically aims to ensure that the main features are not errors and function during development. The manual process frequently causes the publication time of a feature could not to be punctual. This study aims to create an automation script with the data-driven testing method during smoke and regression testing to ensure the quality, especially the main functions or features, can run normally and not be disturbed by the development of new features that are easily understood. One of Various decent automation tools for testing web applications is Katalon Studio which is based on the Selenium tool. The results of the research that have been carried out show that the process of automation of software testing by applying a script automation tool made with Katalon Studio which applies the data-driven testing method is very good with an achievement rate of 80,21%. The automation tools that are built are easy to use, can be learned quickly, are not too complicated and make users want to use it again.
Software testing is considered as one of the most important processes in software development as it checks whether the system meets the requirements and specifications of the users. The testing process manually or automatically aims to ensure that the main features are not errors and function during development. The manual process frequently causes the publication time of a feature could not to be punctual. This study aims to create an automation script with the data-driven testing method during smoke and regression testing to ensure the quality, especially the main functions or features, can run normally and not be disturbed by the development of new features that are easily understood. One of Various decent automation tools for testing web applications is Katalon Studio which is based on the Selenium tool. The results of the research that have been carried out show that the process of automation of software testing by applying a script automation tool made with Katalon Studio which applies the data-driven testing method is very good with an achievement rate of 80,21%. The automation tools that are built are easy to use, can be learned quickly, are not too complicated and make users want to use it again
Improving AI Text Recognition Accuracy with Enhanced OCR For Automated Guided Vehicle
No full textThis artificial intelligence robot uses a mini-computer to operate it and uses mechanical movement like a four-wheeled vehicle with a 2WD drive system. In this article, a control strategy of the AGV robot will be shown and implemented to detect the location. This research Uses OCR (Optical Character Recognition) for the OpenCV library itself which has been enhanced/modified. This enhanced OCR is the main library used in text recognition. This research produces very accurate text detection compared to the default OCR that was previously used on the AGV robot in our university. After the process of reading this text is passed, it will produce text previously read through the camera which will then provide output in the form of text where the AGV robot is located. After the reading is validated, the AGV robot will move to the next point until it returns to its starting point. Based on hardware implementation through testing in the AGV laboratory with artificial intelligence, it can work according to the algorithm and minimize reading errors with a 95% success rate.this artificial intelligence robot uses a mini-computer to operate it and uses mechanical movement like a four-wheeled vehicle with a 2WD drive system. In this article, a control strategy of the AGV robot will be shown and implemented to detect the location. This research Uses OCR (Optical Character Recognition) for the OpenCV library itself which has been enhanced/modified. This enhanced OCR is the main library used in text recognition. This research produces very accurate text detection compared to the default OCR that was previously used on the AGV robot in our university. After the process of reading this text is passed, it will produce text previously read through the camera which will then provide output in the form of text where the AGV robot is located. After the reading is validated, the AGV robot will move to the next point until it returns to its starting point. Based on hardware implementation through testing in the AGV laboratory with artificial intelligence, it can work according to the algorithm and minimize reading errors with a 95% success rate.
 
Digital Forensic on Secure Digital High Capacity using DFRWS Method
Get PDFAs evidenced in the trial, between 2015 and the second quarter of 2022, there were 54 cases involving secure digital high capacity (SDHC) storage hardware as evidenced in trials. In 2021 there will be an increase in cases involving SDHC. The three cases with the highest number are corruption cases, special crimes, and ITE. SDHC is an advanced technology development of Secure Digital (SD) card hardware which functions as storage. SD Card only has a capacity of up to 2 gigabytes, while the largest SDHC capacity is 32 gigabytes. As a storage device that is small, thin, and has a fairly large capacity. this research needs to be done because of the increasingly widespread increase in cases involving SDHC. This study aims to perform digital forensic analysis on SDHC evidence using forensic applications that run on Linux, namely foremost and DC3DD. This study uses the DFRWS method to retrieve valid evidence in court. Based on the research conducted, it was found that the number of files that can be restored at the examination stage using foremost is 77%, and the accuracy of recovered files is 50% with string file hash validation. From this research, it can be concluded that the processing results of DC3DD and Foremost can be used as valid evidence.
As evidenced in the trial, between 2015 and the second quarter of 2022, there were 54 cases involving secure digital high capacity (SDHC) storage hardware as evidenced in trials. In 2021 there will be an increase in cases involving SDHC. The three cases with the highest number are corruption cases, special crimes, and ITE. SDHC is an advanced technology development of Secure Digital (SD) card hardware which functions as storage. SD Card only has a capacity of up to 2 gigabytes, while the largest SDHC capacity is 32 gigabytes. As a storage device that is small, thin, and has a fairly large capacity. this research needs to be done because of the increasingly widespread increase in cases involving SDHC. This study aims to perform digital forensic analysis on SDHC evidence using forensic applications that run on Linux, namely foremost and DC3DD. This study uses the DFRWS method to retrieve valid evidence in court. Based on the research conducted, it was found that the number of files that can be restored at the examination stage using foremost is 77%, and the accuracy of recovered files is 50% with string file hash validation. From this research, it can be concluded that the processing results of DC3DD and Foremost can be used as valid evidence
Classification is one method in image processing. Image processing to search for similar images or with similarity ownership is called image matching or image matching. In the measurement of image matching, the original and fake logo objects are used. Ide
Get PDFThe random noise signal is widely used as a test signal to identify a physical or biological system. In particular, the Gaussian distributed white noise signal (Gaussian White Noise) is popularly used to simulate environmental noise in telecommunications system testing, input noise in testing ADC (Analog to Digital Converter) devices, and testing other digital systems. Random noise signal generation can be done using resistors or diodes. The weakness of the noise generator system using physical components is the statistical distribution. An alternative solution is to use a Pseudo-Random System that can be adjusted for distribution and other statistical parameters. In this study, the implementation of the Gaussian distributed pseudo noise generation algorithm based on the Enhanced Box-Muller method is described. Prototype of noise generation system using a minimum system board based on Cortex Microcontroller or MCU-STM32F4. The test results found that the Enhanced Box-Muller (E Box-Muller) method can be applied to the MCU-STM32F4 efficiently, producing signal noise with Gaussian distribution. The resulting noise signal has an amplitude of ±1Volt, is Gaussian distributed, and has a relatively broad frequency spectrum. The noise signal can be used as a jamming device in a particular frequency band using an Analog modulator.
 
Sentiment Analysis of Beauty Product E-Commerce Using Support Vector Machine Method
Get PDFCustomers who buy goods will provide an assessment in the form of a review. If negative reviews dominate an item, other customers will be reluctant to buy at that store, so customers look for other stores, affecting the store's revenue. Therefore, this study aims to classify e-commerce beauty product reviews using the Support Vector Machine to create a model to categorize beauty product reviews and analyze accuracy. The research phase begins by collecting 50,000 datasets consisting of 35,000 training data and 15,000 test data. After the data is collected, the data labeling stage is carried out, labeled positive and negative. Then the preprocessing step is carried out so that the data is ready to be processed in the feature extraction step. The feature extraction step aims to explore potential information that represents words. Furthermore, the resulting data is evaluated to obtain an accuracy value and determine whether the model made is feasible to use. The results showed that the Support Vector Machine could classify beauty product reviews well with an accuracy of 80.06%.
 
Designing an Ethereum-based Blockchain for Tuition Payment System using Smart Contract Service
Get PDFThe security and disclosure of information in online transaction data remain a sensitive subject until this day. Whenever data collection from a transaction process is accessible over the internet system, certain parties may misuse any one of these data. Blockchain technology, which includes the use of smart contracts, is thought to overcome this problem due to the blockchain's decentralized and distributed nature. Blockchain allows transaction data to be accessed openly and transparently while securely protected by hashing encryption owned by intelligent contracts. This enables users to have detailed access privileges to each transaction's data. The development of smart contracts will be carried out in the production of microservices payment gateways based on decentralized apps (DApps) on the Ethereum blockchain in this research, with the payment gateway generated being used in the tuition payment process. The Truffle framework and the Metamask wallet will be used to assist the Ethereum payment process during the DApps development process. Testing the functionality of each intelligent contract feature reveals that the payment system can be utilized effectively and that there are no issues that cause transaction failures.
Keamanan dan keterbukaan informasi dalam data transaksi yang terjadi secara online masih menjadi suatu permasalahan yang sensitif hingga saat ini. Ketika sekumpulan data dari suatu proses transaksi dapat diakses melalui sistem internet, tidak menutup kemungkinan adanya penyalahgunaan atas setiap data tersebut oleh pihak tertentu. Teknologi blockchain yang mencakup penggunaan smart contract dianggap mampu mengatasi permasalahan ini karena sifat dari blockchain yang terdesentralisasi dan terdistribusi sehingga memungkinkan data transaksi diakses secara terbuka dan transparan, namun setiap data tetap terlindungi dengan aman secara enkripsi hashing yang dimiliki oleh smart contract. Hal tersebut memungkinkan pengguna untuk memiliki hak akses atas setiap data transaksi secara rinci. Pada penelitian ini akan dilakukan pengembangan smart contract dalam pembuatan microservices payment gateway yang berbasis aplikasi terdesentralisasi (DApps) pada ethereum blockchain, dimana payment gateway yang dikembangkan akan digunakan dalam proses pembayaran di suatu universitas. Proses pengembangan DApps akan menggunakan framework Truffle dan wallet Metamask sebagai pendukung adanya proses pembayaran dengan mata uang Ethereum. Hasil pengujian fungsionalitas atas setiap fitur yang dimiliki oleh smart contract menunjukkan bahwa sistem pembayaran dapat digunakan dengan baik dan tidak ditemui permasalahan yang menyebabkan adanya gagal transaks