Future Computer Science Journal

Detection of Cross-Site Scripting Attacks with Code Analysis Using Text Convolution Neural Networks as a Step to Improve User Security

2023-12-24T13:35:50+00:00

The main objective of this research is to build a reliable model capable of classifying Cross-Site Scripting (XSS) attacks through input analysis using the Convolutional Neural Network (CNN) method. The input in question is a JavaScript or HTML script indicated to be included in the XSS attack script. The development of the model architecture is based on the basic Text CNN architecture. The TensorFlow Keras library in Python is used to build the model architecture. The model is trained to study the correlation between data using data from the internet. The model's performance in data classification tasks will be evaluated using accuracy metrics and binary cross-entropy functions. The built model can accurately classify cross-site scripting attack data of 99.95% with a loss rate of 0.29%. To get the optimal model architecture, several experiments are needed to determine the correct number, components, and filter layer size. The Text CNN method for classifying XSS attacks is a new approach to detecting and preventing XSS attacks. The proposed CNN method is specifically for text processing that has been widely used in various fields and has proven performance. Input analysis is the foremost approach used and is crucial in preventing XSS attacks, considering that these attacks are generally carried out by code injection. The built model can accurately classify cross-site scripting attack data of 99.95% with a loss rate of 0.29%. The application of the Text CNN method makes the proposed model quite reliable and able to outperform previous methods in the XSS attack classification task.

Text Normalization on Indonesian-English Code-Mixed Twitter Text using UFAL ByT5

2023-12-22T00:51:47+00:00

Social media has been grown rapidly in the global community. It also includes Twitter, which is getting increase in both users and content created. However, Twitter has character limit in one tweet which causes changes to the writing patterns of its users. Twitter users began to modify their writing from using formal words into non-formal words, one of which was using code-mixed language. For tweet analysis purposes, text normalization is required to transform non-formal words into formal ones to help analysis process. The recent state-of-the-art for Indonesian-English code-mixed Twitter text normalization is with statistical machine translation (SMT) models, however the SMT model still has weakness in word recognition. This research focuses on the Indonesian and English code-mixed Twitter text normalization using one of transformer model which is UFAL ByT5. There are two UFAL ByT5 models that were used, each of them are for Indonesian and English language. Research result shows that UFAL ByT5 model outperform SMT model on text normalization by 0.88 percent of BLEU score in difference.

Monitoring Smartfarm Using IoT Based for Rice Agriculture

2023-12-24T13:34:50+00:00

Indonesia is an agrarian country where the majority of its population chooses farming as their occupation, especially for those living in village areas. The most crucial factor influencing agricultural outcomes is the quality of farmland, which depends on environmental conditions such as soil moisture, humidity, and temperature in the farmland itself. These environmental factors are affected by the seasonal changes in Indonesia, namely the rainy season, which provides abundant water for plant energy, and the dry season, which has limited and irregular water supply. The implementation of technology is expected to help the agricultural sector withstand climate change and improve agricultural productivity, thereby increasing farmers income. This research utilizes technology for monitoring agricultural land, particularly focusing on soil moisture, humidity, temperature, and water levels in rice fields. The smart farm monitoring system can assist farmers in monitoring the condition of agricultural land, with criteria including soil moisture, humidity, temperature, and water levels. The smart farm monitoring system is designed to connect to the Internet of Things (IoT), where the monitoring system sends data on soil moisture, air humidity, air temperature, and water levels detected through devices. The detected data is then transmitted to the smart farm web, accessible through smartphones or laptops, allowing remote monitoring.The research results indicate a success rate of 95.9% for soil moisture sensors, 98.6% for ultrasonic sensors, 97.3% for humidity measurements, and 95.3% for temperature measurements. This translates to: Which means that the research to develop this monitoring tool was successful, as evidenced by the high success rate in its experimental use.

Classifying Empathy in Textual Anamnesis using Single Layered LSTM Performance

2023-12-18T09:45:53+00:00

This study explores the challenges of integrating artificial intelligence (AI) in healthcare 5.0, particularly the perceived lack of empathy in AI applications and its potential impact on human connections in healthcare services. Patient-centered healthcare relies on effective communication, and the paper highlights the absence of computational studies detecting empathetic support in clinical interviews. The research distinguishes between cognitive and affective empathy, noting the medical emphasis on cognitive empathy for objectivity and detachment, while cautioning against affective empathy due to concerns about emotional strain. The proposed solution involves employing Long Short-Term Memory (LSTM) for sentiment analysis in clinical conversations, citing its accuracy and efficiency. Methodologically, the study annotates conversations for affective and cognitive empathy, performs word embedding using word2vec and GloVe, and trains the LSTM model on a dataset of medical interviews. Results suggest promising outcomes in classifying empathy levels, though limitations are acknowledged, such as the simplified empathy aspects and the need for more robust validation methods. The study contributes to the development of Vidya Medic, an AI-assisted intelligent healthcare system, with potential implications for enhancing empathetic communication in clinical settings. This experiment's findings suggest that the LSM model has been developed for the purpose of classifying empathy levels in clinical conversations. However, there are limitations to this study, including the simplification of empathy aspects, which focus solely on affective and cognitive empathy

Flood Disaster Detection Based on Rainfall using Random Forest Algorithm

2023-12-22T00:51:13+00:00

This research was conducted to detect worldwide natural disasters, namely floods based on rainfall, using the Random Forest algorithm. This research can also help the government estimate what will be done in the future when a flood disaster occurs. The method is to collect rainfall and flood disaster data from Kaggle. The data is then cleaned, processed, and tested for reliability. Using a suitable dataset for testing and choosing the correct algorithm can ensure that the data mining process produces accurate information. Furthermore, the Random Forest algorithm was applied to the data to classify flood disasters based on rainfall. The results showed that the Random Forest algorithm could provide flood disaster classification results with a high accuracy rate of 95.8%. The novelty of this research lies in using the Random Forest algorithm, which is rarely used in flood disaster classification research based on rainfall. This research is expected to contribute to developing a more accurate and reliable rainfall-based flood disaster early warning system.