Jurnal Teknologi dan Sistem Komputer

Journal Information
ISSN / EISSN : 2620-4002 / 2338-0403
Published by: Diponegoro University (10.14710)
Total articles ≅ 341
Current Coverage
Archived in

Latest articles in this journal

Fauzi Ihsan, Iwan Iskandar, Nazruddin Safaat Harahap, Surya Agustian
Jurnal Teknologi dan Sistem Komputer, Volume 9, pp 199-204; https://doi.org/10.14710/jtsiskom.2021.13907

Hate speech and abusive language are easily found in written communications in social media like Twitter. They often cause a dispute between parties, the victims, and the first who write the tweet. However, it is also difficult to distinguish whether a tweet contains hate speech and/or abusive language for those who take sides. This research aims to develop a method to classify the tweets into abusive and/or contain hate speech classes. If hate speech is detected, then the system will measure the hardness level of hatred. The dataset includes 13,126 real tweets data. Word embeddings are used for featuring text input. For the tweets classification, we use a Decision Tree algorithm. Some engineering of features and parameters tuning has improved the classification of the three classes: hate speech class, abusive words, and hate speech level. The lexicon feature in the Decision Tree classification produces the highest accuracy for detecting the three classes rather than engineering special features and textual features. The average accuracy of the three classes increased from 69.77 % to 70.48 % for the training-testing composition of 90:10, and another 69.35 % to 69.54 % for 80:20 respectively.
, Rifki Adhitama, Alon Jala Tirta Segara
Jurnal Teknologi dan Sistem Komputer, Volume 9, pp 191-198; https://doi.org/10.14710/jtsiskom.2021.13970

Currency recognition is one of the essential things since everyone in any country must know money. Therefore, computer vision has been developed to recognize currency. One of the currency recognition uses the SIFT algorithm. The recognition results are very accurate, but the processing takes a considerable amount of time, making it impossible to run for real-time data such as video. AKAZE algorithm has been developed for real-time data processing because of its fast computation time to process video data frames. This study proposes the faster real-time currency recognition system on video using the AKAZE algorithm. The purpose of this study is to compare the SIFT and AKAZE algorithms related to a real-time video data processing to determine the value of F1 and its speed. Based on the experimental results, the AKAZE algorithm is resulting F1 value of 0.97, and the processing speed on each video frame is 0.251 seconds. Then at the same video resolution, the SIFT algorithm results in an F1 value of 0.65 and a speed of 0.305 seconds to process one frame. These results show that the AKAZE algorithm is faster and more accurate in processing video data.
Akinbowale Nathaniel Babatunde, Afeez Adeshina Oke, Abdulkareem Ayopo Oloyede, Aisha Oiza Bello
Jurnal Teknologi dan Sistem Komputer, Volume 9, pp 205-210; https://doi.org/10.14710/jtsiskom.2021.14038

This paper aims to improve the image scrambling and encryption effect in traditional two-dimensional discrete Arnold transform by introducing a new Residue number system (RNS) with three moduli and the New Arnold Transform. The study focuses on improving the classical discrete Arnold transform with quasi-affine properties, applying image scrambling and encryption research. The design of the method is explicit to three moduli set {2n, 2n+1+1, 2n+1-1}. These moduli set includes equalized and shapely moduli leading to the effective execution of the residue to binary converter. The study employs an arithmetic residue to the binary converter and an improved Arnold transformation algorithm. The encryption process uses MATLAB to accept a digital image input and subsequently convert the image into an RNS representation. The images are connected as a group. The resulting encrypted image uses the Arnold transformation algorithm. The encrypted image is used as input at decryption using the anti-Arnold (Reverse Arnold) transformation algorithm to convert the picture to the original RNS (original pixel value). Then the RNS was used to retransform the original RNS to its binary form. Security analysis tests, like histogram analysis, keyspace, key sensitivity, and correlation coefficient analysis, were administered on the encrypted image. Results show that the hybrid system can use the improved Arnold transform algorithm with better security and no constraint on image width and size.
Jurnal Teknologi dan Sistem Komputer, Volume 9, pp 211-217; https://doi.org/10.14710/jtsiskom.2021.14197

Osteoporosis is one type of disease that is not easily detected. This disease can cause fractures for the sufferer. Early detection of osteoporosis is crucial to prevent fractures. This study aims to detect osteoporosis through features extracted from cortical bone and trabeculae in dental panoramic images. The results of the selected feature extraction are trained using an artificial neural network. Based on the study results, the dominant features for osteoporosis detection are radio morphometric index and morphological features. The accuracy, sensitivity, and specificity of the J48 and Learning Vector Quantization (LVQ) are 83.88 %, 78.57 %, and 100 %, respectively.
, Muiz Olalekan Raheem, Muyideen Abdulraheem, Idowu Dauda Oladipo, , Omotola Fatimah Baker
Jurnal Teknologi dan Sistem Komputer, Volume 9, pp 224-229; https://doi.org/10.14710/jtsiskom.2021.13965

As a result of advancements in technology and technological devices, data is now spawned at an infinite rate, emanating from a vast array of networks, devices, and daily operations like credit card transactions and mobile phones. Datastream entails sequential and real-time continuous data in the inform of evolving stream. However, the traditional machine learning approach is characterized by a batch learning model. Labeled training data are given apriori to train a model based on some machine learning algorithms. This technique necessitates the entire training sample to be readily accessible before the learning process. The training procedure is mainly done offline in this setting due to the high training cost. Consequently, the traditional batch learning technique suffers severe drawbacks, such as poor scalability for real-time phishing websites detection. The model mostly requires re-training from scratch using new training samples. This paper presents the application of streaming algorithms for detecting malicious URLs based on selected online learners: Hoeffding Tree (HT), Naïve Bayes (NB), and Ozabag. Ozabag produced promising results in terms of accuracy, Kappa and Kappa Temp on the dataset with large samples while HT and NB have the least prediction time with comparable accuracy and Kappa with Ozabag algorithm for the real-time detection of phishing websites.
Vega Purwayoga
Jurnal Teknologi dan Sistem Komputer, Volume 9, pp 167-173; https://doi.org/10.14710/jtsiskom.2021.14003

The distribution of personal protective equipment (PPE) plays a vital role in meeting the needs of PPE in an area. This study aims to measure the priority of PPE recipient regions in West Java Province using a skyline query algorithm, namely Sort Filter Skyline (SFS). In this study, the SFS algorithm is modified to optimize the dominance measurement section. Regions that do not have hospitals will not be prioritized for PPE recipients. The preferences used in this study are maximum and minimum. The maximum preference rule is used for the number of ODP, PDP, positive and dead cases, while the minimum preference rule is used for the cured and distance attributes. The application of SFS for calculating priority regions has been successfully carried out by developing two models, namely MS1 using unmodified SFS and MS2 using modified SFS by adding a selection process for regions with no hospitals. The MS1 produces 21 skyline objects (55.55 %), while MS2 15 (66.66 %) skyline objects. The MS2 is faster than that of MS1 because fewer objects are being tested. The MS1 takes 0.0222 seconds, while MS2 only 0.0193 seconds.
Moh. Arie Hasan, Yan Riyanto, Dwiza Riana
Jurnal Teknologi dan Sistem Komputer, Volume 9, pp 218-223; https://doi.org/10.14710/jtsiskom.2021.14013

This study aims to classify the disease image on grape leaves using image processing. The segmentation uses the k-means clustering algorithm, the feature extraction process uses the VGG16 transfer learning technique, and the classification uses CNN. The dataset is from Kaggle of 4000 grape leaf images for four classes: leaves with black measles, leaf spot, healthy leaf, and blight. Google images of 100 pieces were also used as test data outside the dataset. The accuracy of the CNN model training is 99.50 %. The classification yields an accuracy of 97.25 % using the test data, while using test image data outside the dataset obtains an accuracy of 95 %. The designed image processing method can be applied to identify and classify disease images on grape leaves.
Sesar Prabu Dwi Sriyanto, Ping Astony Angmalisang, Lusia Manu, Joshian N. W. Schaduw, Calvyn F. A. Sondak, Rose O. S. E Mantiri, Alfret Luasunaung, Deiske A. Sumilat
Jurnal Teknologi dan Sistem Komputer, Volume 9, pp 180-190; https://doi.org/10.14710/jtsiskom.2021.14009

The automatic tsunami detection algorithm needs to be put in the sea level observation system to give society a quick warning when a tsunami happens. This study designs an automatic tsunami detection algorithm consisting of three sub-algorithm: spike elimination, gap data filling, and tsunami detection. Spike elimination and gap data filling are used to improve the sea level data, which is often disturbed by spikes and gap data due to electronic factors. This algorithm was tested using time-series tide gauge data that contain tsunami waveforms in Indonesia from 2007 to 2019. About 54.52 % of 409 spikes have been eliminated while the gap data were successfully filled. Furthermore, tsunami detection, which uses DART (Deep-ocean Assessment and Reporting of Tsunamis) and TEDA (Tsunami Early Detection Algorithm) methods, can detect 7 of 10 tsunami waveforms. However, there are three undetected tsunamis and one false detection. This algorithm has an average delay of 7.7 minutes in detection time.
Maya Fitria, Cosmin Adrian Morariu, , Ramzi Adriman
Jurnal Teknologi dan Sistem Komputer, Volume 9, pp 174-179; https://doi.org/10.14710/jtsiskom.2021.14125

It is necessary to conserve important information, like edges, details, and textures, in CT aortic dissection images, as this helps the radiologist examine and diagnose the disease. Hence, a less noisy image is required to support medical experts in performing better diagnoses. In this work, the non-local means (NLM) method is conducted to minimize the noise in CT images of aortic dissection patients as a preprocessing step to produce accurate aortic segmentation results. The method is implemented in an existing segmentation system using six different kernel functions, and the evaluation is done by assessing DSC, precision, and recall of segmentation results. Furthermore, the visual quality of denoised images is also taken into account to be determined. Besides, a comparative analysis between NLM and other denoising methods is done in this experiment. The results showed that NLM yields encouraging segmentation results, even though the visualization of denoised images is unacceptable. Applying the NLM algorithm with the flat function provides the highest DSC, precision, and recall values of 0.937101, 0.954835, and 0.920517 consecutively.
Arif Amrulloh,
Jurnal Teknologi dan Sistem Komputer, Volume 9, pp 157-166; https://doi.org/10.14710/jtsiskom.2021.14137

Scheduling courses in higher education often face problems, such as the clashes of teachers' schedules, rooms, and students' schedules. This study proposes course scheduling optimization using genetic algorithms and taboo search. The genetic algorithm produces the best generation of chromosomes composed of lecturer, day, and hour genes. The Tabu search method is used for the lecture rooms division. Scheduling is carried out for the Informatics faculty with four study programs, 65 lecturers, 93 courses, 265 lecturer assignments, and 65 classes. The process of generating 265 schedules took 561 seconds without any scheduling clashes. The genetic algorithms and taboo searches can process quite many course schedules faster than the manual method.
Back to Top Top