Stunting is a growth disorder in toddlers characterized by a height below the age-standard due to chronic nutritional deficiencies. This study aims to design and compare prediction models for stunting levels using two Fuzzy Inference System (FIS) methods: Tsukamoto and Mamdani. The data used comes from the secondary dataset "data set clustering gizi" obtained from Kaggle, consisting of 495 todd…
Music plays a significant role through digital streaming platforms, while concerns about offensive language in lyrics, especially among young listeners, require attention. This study develops a BERT-based model to detect offensive language in English song lyrics. The dataset comprises 3,598 lyrics, divided into training 2,878 data, validation 360 data, and test 360 data. Experiments tested 12 p…
Traditional cake image classification aims to recognize various types of cakes based on visual features, thereby assisting in the effort to introduce traditional cakes. Indonesian traditional cakes hold significant cultural value that must be preserved while also presenting substantial potential in supporting tourism through culinary tourism. This study develops a classification model based on …
This study aims to optimize the process of selecting football players in the PES 2017 game using genetic algorithms. The optimization is conducted to identify the best combination of players based on the average player ratings over a season, thereby maximizing team performance in the game. The methodology employed in this research is experimental, utilizing the application of genetic algorithms…
Classification is one of the fields of Natural Language Processing that automates the classification of text into one or more appropriate categories based on its content by building a model using training data. This study aims to classify hate speech tweets on application X using Random Forest and Synthetic Minority Oversampling Technique (SMOTE). The Random Forest approach is used because of i…
The increasing use of social media has led to significant growth in Indonesian text data, creating complexity in emotion identification and classification tasks. To address this challenge, this study develops an emotion classification system using the fine-tuning method on the IndoBERT model. This research aims to classify emotions in Indonesian text using IndoBERT fine-tuning. The study utiliz…
The publication rate of scientific articles has significantly increased over time. This presents a challenge for journal administrators and academics in organizing and sorting these articles to align with the journal's scope. This study aims to address this issue by developing a scientific article clustering system utilizing MultiBERT as the data representation model and K-Means for cluster ide…
The increasing number of multidisciplinary undergraduate theses in Informatics field presents challenges in categorizing these documents accurately. This study develops a system to measure the semantic similarity of final project documents to the ontology of Informatics using the Wu Palmer method as a solution to this problem. This method utilizes ontology tree structures and taxonomic depth to…
Tuberculosis is still one of the infectious diseases that causes high mortality rates worldwide. Early detection using Chest X-Ray images is very important in the effort to treat this disease. However, manual interpretation of X-ray results performed by doctors can cause differences in diagnosis. This study aims to classify tuberculosis in Chest X-Ray images by implementing three Convolutional …
This study aims to develop and evaluate a Topic Modelling and Topic Generation system using user review data from the Steam platform. The research utilizes the BERTopic model to cluster reviews into topics in an unsupervised manner, and the Qwen2.5 Large Language Model to generate more informative topic representations. Various algorithm combinations covering embedding, dimensionality reduction…
A sentence can contain various named entities with important meanings, such as names of people, locations, organizations, and time expressions. However, extracting this information manually requires significant time and resources. Named Entity Recognition (NER) offers an automated solution that improves the efficiency of this task. One method for developing an NER system is by using BERT, a tra…
TikTok Shop, as a social commerce feature in the TikTok application, has become a popular shopping platform in Indonesia. However, government regulations prohibiting direct transactions on social media forced TikTok to stop this service from October 4 until it reopened on December 12, 2023, after establishing a partnership with Tokopedia. This incident triggered various public opinions. This se…
Prediction or also called forecasting is the process of estimating events that will occur in the future. In this research, software is developed that can predict crime rates using the Long Short-term Memory method using the Los Angles crime dataset which has a crime type domain of 139 with details of 9 types of crimes that have the most frequency, namely; vehicle - stolen, battery - simple assa…
Kidney stones are one of the most common health problems and have the potential to cause chronic kidney failure if not properly addressed. To support early prevention efforts, this study aims to compare two fuzzy inference methods, namely Mamdani and Tsukamoto. The study will use 89 data points obtained from Kaggle under the title "Kidney Stones Dataset." In predicting the risk of kidney stones…
In this global era, the demand for coffee beans has been increasing year by year. The economic value of coffee beans is greatly influenced by the quality of the beans being marketed. Manual assessment of coffee quality by experts or farmers tends to be subjective and time-consuming. Based on these conditions, this study developed software to detect coffee bean quality in real-time using the YOL…
Klasifikasi citra makanan merupakan tantangan penting dalam computer vision dengan aplikasi luas pada sistem pencatatan gizi dan pelacakan kalori, namun penelitian yang secara langsung membandingkan arsitektur CNN dan Transformer pada dataset makanan multikelas dengan variasi visual tinggi masih terbatas. Penelitian ini membandingkan dua arsitektur deep learning, yaitu DenseNet121 dan Vision Tr…
The rice farming sector plays a strategic role in supporting national food security; however, its sustainability is under serious pressure due to the conversion of paddy fields into oil palm plantations. In Banyuasin Regency, the sustainability of rice farming has been increasingly threatened by this land-use conversion. The binary logistic regression analysis shows that farm income is the only…
This study aims to develop an interactive dialogue system based on Speech Recognition in games, allowing players to interact with Non-Player Characters (NPCs) through real-time voice commands. The method used is Fuzzy String Matching with the Levenshtein Distance algorithm to calculate the similarity between speech-to-text conversion results and predefined system commands. The research stages i…
Advances in computer vision and artificial intelligence technology are driving innovation in human computer interaction, one example of which is Hand Gesture Recognition. In maze games, control systems generally still use conventional devices such as keyboards, mice, or joysticks, thus not providing a fully natural and interactive gaming experience. Therefore, this study implemented a Hand Gest…
The increasing spread of hate speech on social media, particularly on Twitter (X), poses challenges for manual content moderation, which tends to be ineffective. This condition highlights the need for an automated classification system capable of accurately detecting hate speech. This study aims to develop a hate speech classification model using approach combining IndoBERT and Convolutional Ne…
Game application reviews, such as those for Roblox, often exhibit significant class imbalance, where positively labeled reviews vastly outnumber negative ones. This study aims to compare the effectiveness of Random Majority Under-sampling (RMU) and Random Over-sampling (ROS) data balancing techniques on the Bidirectional Encoder Representations from Transformers (BERT) architecture for automati…
The rapid development of digital technology has increased the spread of online gambling promotional content across various digital platforms, making manual moderation less effective. This study aims to develop an online gambling text classification model using Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM). CNN was selected for its ability to recognize important word patte…
Pneumonia is a lung infection with a high mortality rate, ranking as the fourth leading cause of global deaths in 2020 according to the WHO. In this study, Chest X-Ray images were classified into pneumonia and normal categories using the Convolutional Neural Network (CNN) method with EfficientNetV2-B0, EfficientNetV2-B1, EfficientNetV2-B2, and EfficientNetV2-B3 architectures, which were subsequ…
The banking industry faces significant challenges in credit risk management due to the subjectivity of manual analysis. This study aims to develop an objective credit risk classification model using Support Vector Machine (SVM) optimized with Grid Search (GS) and Synthetic Minority Over-sampling Technique (SMOTE) on the German Credit Dataset. Recall is designated as the primary evaluation metri…
This study aims to design and develop a web-based Student Achievement Information System (SIPRESMA) for the Faculty of Computer Science, Sriwijaya University. The management of student achievement data is still carried out manually, resulting in less effective processes for recording, validating, searching, and reporting achievement data. To address these issues, a system was developed to integ…
Heart disease is one of the leading causes of death worldwide, making accurate early detection methods essential. Machine learning can be used to classify patients conditions, but data imbalance and feature selection can affect model performance. This study applied Chi-Square feature selection and the Synthetic Minority Over-sampling Technique (SMOTE) to 918 patient records with 11 clinical att…
Eczema is a skin condition that can be triggered by food allergens, requiring sufferers to carefully examine the composition of packaged food products. The complexity of ingredient labels often makes manual identification difficult. This study aims to develop a food composition classification system to detect allergens that trigger eczema using the Convolutional Recurrent Neural Network (CRNN) …
The success of a game no longer depends solely on its technical functionality, but also on its ability to establish player engagement. This study aims to develop Disgaster 2 as a gas leak disaster simulation game, as well as to measure the game engagement experience using the GUESS-18. The game was developed using the Unity 2022 game engine with the GDLC methodology. Testing was conducted throu…
Sentiment analysis is one of the applications of Natural Language Processing (NLP) used to identify users’ opinions or evaluations toward a particular service or object. This study aims to develop a sentiment analysis system for Tripadvisor reviews using the Transformer-based XLM-RoBERTa model. The dataset used consists of 28.754 Italian-language Tripadvisor reviews classified into two sentim…