JTIM : Jurnal Teknologi Informasi dan Multimedia
Not a member yet
290 research outputs found
Sort by
Comparison of Social Media Between Tiktok and Instagram to Detect Negative Content Using Natural Language Processing Method
No full textIn the digital era, social media platforms have become essential tools for communication, content creation, and information dissemination. However, with the increasing volume of user-generated content, the spread of negative or harmful content has emerged as a major challenge for platform administrators and users alike. This study aims to compare TikTok and Instagram in their capacity to detect and manage negative content using Natural Language Processing (NLP) techniques. A dataset of 2,000 user comments was collected—1,000 from each platform—through web scraping. These comments were analyzed using a variety of NLP methods, including sentiment analysis tools (VADER and TextBlob), text classification algorithms (Support Vector Machine and Random Forest), and Named Entity Recognition (NER) using the spaCy library. The comparison was conducted based on the classification performance of each NLP technique in detecting negative content, considering metrics such as accuracy, precision, recall, and F1-score. The results showed that while both SVM and Random Forest performed well in classification tasks, SVM outperformed the others in terms of overall accuracy and consistency across platforms. Sentiment analysis provided a general overview of content polarity, but it was less effective in detecting nuanced or sarcastic language. NER contributed to identifying specific entities that may be associated with negative expressions, enriching the contextual understanding of comments. This study highlights the potential of combining multiple NLP methods to improve automated content moderation systems. It also underlines the importance of platform-specific characteristics, such as user behavior and engagement style, which influence the nature and frequency of negative content. Future work should focus on improving the handling of contextual ambiguity and sarcasm to ensure more robust and adaptive moderation technologies across different social media platforms
Implementasi Arsitektur Deep Convolutional Neural Network (CNN) dengan Transfer Learning untuk Klasifikasi Penyakit Kulit
No full textSkin diseases are common health problems that require early diagnosis to prevent serious complications. This study aims to develop an automatic skin disease image classification system using a transfer learning approach based on Convolutional Neural Networks (CNN). Image datasets were obtained from Kaggle and underwent preprocessing stages including resizing, normalization, and augmentation. Four CNN architectures were evaluated: VGG16, ResNet50, MobileNetV2, and InceptionV3, implemented using Python and the Keras library on the Google Colab platform. The dataset was split into three training and testing ratios (90:10, 80:20, and 70:30) to assess the impact of data proportion on model performance. Models were trained by modifying the output layer to match the number of classes, and evaluated using accuracy, precision, recall, F1-score, confusion matrix, and ROC curve metrics. The results show that a 70:30 ratio yielded the most optimal training performance. InceptionV3 achieved the highest validation accuracy at 80.04%, but experienced overfitting, while VGG16 demonstrated better generalization to test data. This study proves that transfer learning with CNN is effective in improving the accuracy of automatic skin disease diagnosis and has the potential to become an efficient diagnostic solution, especially in areas with limited medical infrastructure
Pengaruh Teknik Representasi Teks Bag-of-Words dan TF-IDF terhadap Akurasi Klasifikasi Sentimen Teks Multi-Domain
No full textRepresentasi teks merupakan komponen esensial dalam sistem analisis sentimen, karena menentukan bagaimana data teks diubah menjadi fitur numerik yang dapat dimanfaatkan oleh algoritma klasifikasi. Penelitian ini bertujuan untuk menganalisis pengaruh dua teknik representasi teks populer, yaitu Bag-of-Words (BoW) dan Term Frequency–Inverse Document Frequency (TF-IDF), terhadap performa klasifikasi sentimen teks pendek dalam konteks multi-domain. Dataset yang digunakan merupakan hasil kombinasi antara data asli dan data augmentasi berbasis sinonim, dengan total 418 entri teks. Dua algoritma pembelajaran mesin yang digunakan dalam evaluasi adalah Ridge Classifier dan Complement Naïve Bayes. Penilaian dilakukan menggunakan teknik validasi silang Stratified K-Fold serta empat metrik evaluasi utama: akurasi, presisi, recall, dan F1-score. Hasil eksperimen menunjukkan bahwa representasi TF-IDF secara konsisten memberikan performa lebih baik dibandingkan BoW pada kedua model. Konfigurasi terbaik dicapai oleh Ridge Classifier dengan TF-IDF, yang memperoleh akurasi sebesar 0,911 dan F1-score sebesar 0,908. Temuan ini menggarisbawahi pentingnya pemilihan teknik representasi fitur yang tepat dalam meningkatkan efektivitas sistem klasifikasi sentimen berbasis teks
Perencanaan Pengembangan Sistem Informasi Penjadwalan Kuliah Praktikum di STMIK AMIKOM Yogyakarta
No full textPracticum course scheduling is a complex task in higher education institutions as it involves multiple parameters such as lecturer availability, room capacity, and time slots. This process poses a significant challenge for laboratory administrators in ensuring that scheduling conflicts are avoided and that all resources are utilized optimally. This study implements a Genetic Algorithm (GA) to optimize the practicum course scheduling process at STMIK AMIKOM Yogyakarta, which has since been renamed Universitas Amikom Yogyakarta. The methodological stages include population initialization, fitness evaluation, selection using the Roulette Wheel Selection method, crossover using One Point Crossover, and mutation using Targeted Mutation. The results demonstrate that the genetic algorithm successfully produces optimal solutions by eliminating lecturer and room conflicts, while also maximizing equitable time utilization. During the iteration phase, the algorithm generated a conflict-free practicum schedule with a maximum fitness value of 167. The process terminated at the first generation after identifying two optimal chromosomes out of ten. These findings confirm that the genetic algorithm is effective in solving practicum scheduling problems and can be applied to minimize schedule clashes and improve operational efficiency in academic environments
Implementasi Augmented Reality Sebagai Media Pembelajaran Untuk Pengenalan Buah-Buahan
No full textInnovative and interactive teaching strategies have emerged as a result of the development of communication and information technology. One of the most promising and rapidly expanding educational technologies is augmented reality. By displaying virtual things in three dimensions in a real-world setting in real time, Augmented Reality can make studying more engaging and joyful for students Augmented reality can display virtual objects in three dimensions in real time, creating a more engaging and enjoyable learning experience for students. This research aims to develop and implement Augmented Reality-based fruit recognition learning media as an alternative to conventional, static and unengaging media for elementary school students, helping them visualize the concepts being learned. The Multimedia Development Life Cycle (MDLC), which has six stages concept, design, gathering materials, assembly, testing, and distribution the development methodology. This application is designed to display various types of fruit as 3D objects that can be scanned through markers using the camera on an android device. Each fruit is equipped with its own name and information to improve student knowledge. Testing is carried out through black box testing to evaluate system functions, and user feasibility testing using a Likert scale questionnaire given to 15 grade students. According to the results of black-box testing, there were no system or functional issues and the application operated as planned. It received an 84.86% feasibility score, placing it in the "very feasible" range. Thus, it can be said that this AR-based fruit recognition app works well to boost students\u27 curiosity, involvement, and comprehension of the material
Pengembangan Back-end pada Aplikasi Smart Nutrition Berbasis Node.js dan Hapi dengan Integrasi Google Cloud Platform
No full textAdvances in digital technology drive the need for smart and integrated nutrition monitoring systems, but developers often focus only on features without considering architectural design. This research aims to develop and implement a RESTful API on the Google Cloud Platform (GCP) backend for the Smart Nutrition App, which has the ability to support daily fruit consumption tracking powered by machine learning. The methodology used is based on the Software Development Life Cycle (SDLC) model, including requirements analysis, cloud-native system design, modular API development using Node.js and Hapi.js, functional testing in Postman, and stress testing in K6 to 4000 virtual users. The results show that the RESTful API can sustain a load of up to 1000 virtual users with 0% error rate, but performance degrades very sharply above this level, to the point where the error rate is 100% at 4000 users. These findings indicate the need for infrastructure optimization to support the demands of real applications. The result of this research is that the system meets the functional requirements and performs well at small scale but requires infrastructure improvements such as load balancing and auto-scaling for scaled environments. The main contribution of this research is to present a scalable and modular backend framework for Smart Nutrition App as a future reference when developing similar systems
Evaluasi Pengguna Hospital Management Information System (HMIS) dengan Metode End User Computing Satisfaction (EUCS) di Instalasi Rawat Jalan RSUP Dr. Kariadi Semarang
No full textThe Hospital Management Information System (HMIS) is basically used in various service pro-cesses starting from patient registration to the payment process which is expected to streamline the performance of the hospital and provide credible information for its users. As technology develops, HMIS has been used in various hospitals, one of which is to support all aspects of in-creasing customer satisfaction. RSUP Dr. Kariadi Semarang has Type A accreditation status which is a teaching and referral hospital in Central Java. Seeing the high activity at RSUP Dr. Kariadi, customer satisfaction is one of the references. Therefore this study focuses on evaluating consumer satisfaction using the End User Computing Satisfaction (EUCS) evaluation model through direct interviews. The four variables used include Content, Accuracy, Format, Ease of Use, and Timelines. Based on the results of the analysis, there are 3 dimensional factors that have fulfilled user satisfaction, namely the dimensions of content, accuracy, and format. Even though it is stated that it meets user satisfaction, these dimensions still need to be improved because they still have weaknesses in terms of multiple patient data collection and identification of new patients and old patients. The other two dimensions, namely the dimensions of timeliness and ease of use, are still not sufficient for user satisfaction because of the capabilities of the devices used and the lack of guidance in the system
Comparative Analysis of Stock Price Prediction Using Deep Learning with Data Scaling Method
No full textThe dynamic and unpredictable nature of stock prices makes accurate forecasting an important challenge in financial analysis. This study aims to compare the performance of three deep learning models, namely, Recurrent Neural Network (RNN), Gated Recurrent Unit (GRU), and Long Short-Term Memory (LSTM) in predicting stock prices on historical daily banking data from Yahoo Finance. The main objective is to determine the model that is best able to capture sequential patterns and temporal dependencies in stock price movements. Each model was trained and op-timized through data scaling, namely MinMax Scaler and Standard Scaler, with performance evaluated using Root Mean Square Error (RMSE) as the primary metric. Results show that while the RNN provides a basic approach, the GRU and LSTM models produce higher prediction accuracy, with GRU achieving the lowest RMSE thanks to its better ability to maintain long-term depend-encies. The RMSE achieved by RNN, GRU, and LSTM were 211.47, 158.89, and 197.45, respectively. The lowest error results were achieved when using MinMax Scaler. The use of MinMax Scaler here shows a better performance improvement with an average improvement of 22.57% compared to using Standard Scaler. This comparative analysis contributes to providing empirical insight into the relative effectiveness of the tested architectures. The findings suggest that the combination of GRU and MinMax Scaler can be a more reliable tool for financial forecasting, with the potential to develop more robust stock prediction applications under fluctuating market conditions
Optimalisasi Potensi Wisata NTT dari Perspektif Google Trends dan Big Data Analytics
No full textThis study aims to identify and analyze tourism trends in East Nusa Tenggara (NTT) using the K-Means clustering method integrated with Google Trends and Big Data Analytics. By utilizing data that includes the number of tourist attractions, hotel accommodations, tourist visits (Domestic and foreign), and restaurant accomodation, the NTT region is categorized into several clusters based on tourism characteristics. The analysis results reveal three main clusters: areas with low tourist attractions and accommodations, areas with very high tourist attractions, and areas with good accommodation facilities but moderate attractions. These findings provide crucial insights for policymakers and tourism industry stakeholders to formulate more effective development strategies, such as infrastructure enhancement in high-potential areas and targeted promotion for niche markets. Additionally, the analysis results indicate significant fluctuations in tourist interest towards NTT, with peak searches occurring in April and September. This research utilizes data from Google Trends and other sources to analyze trends and tourist attractions in NTT tourism, thereby aiding in the development of more effective promotional strategies. Overall, this study contributes to a deeper understanding of tourism dynamics in NTT and the necessary optimization steps to enhance the competitiveness of these destinations. With this data-driven approach, it is hoped that the tourism sector in NTT can develop sustainably and provide economic benefits to local communities
Perencanaan Arsitektur Enterprise SI/TI pada SMA Negeri 1 Purwoasri Kediri menggunakan Kerangka TOGAF
No full textDigital transformation in the education sector requires institutions such as SMA Negeri 1 Purwoasri Kediri to adopt a structured and strategic approach in managing Information Systems and Information Technology (IS/IT). Fragmented systems, duplicated data, and inefficient services have become major challenges that hinder optimal performance and service delivery. This study aims to design an Enterprise Architecture (EA) using the TOGAF Architecture Development Method (ADM) framework to address these issues comprehensively. The research method combines observation, Focus Group Discussions (FGDs), and a structured analysis through TOGAF ADM phases—Preliminary, Architecture Vision, Business Architecture, Information Systems Architecture, and Technology Architecture. An evaluation of the institution’s current IS/IT condition was conducted using the EA Capability Maturity Model (EA-CMM) Scorecard. The results indicate that the school has a relatively strong foundation in IT governance (level 4/Managed), but still shows limitations in methodology usage, supporting tools, and human resource competencies, with an average score of 2.71 (Under Development). Business process mapping identified several priority systems, including student admission (PPDB), e-report cards, and digital libraries, alongside supporting systems such as inventory and asset management. The overall system effectiveness rate is 72.5%, these findings align with previous studies highlighting the urgency of integrated systems in schools. The study provides practical implications, including recommendations for phased implementation of EA, continuous training for human resources, and the development of an embedded EA model tailored to the needs and conditions of secondary educational institutions, ensuring sustainability and adaptability in the long term