ComniTech : Journal of Computational Intelligence and Informatics

Sentiment Analysis of Hotel Guests Using the Naive Bayes Classifier

Ahmad Zamsuri, Tengku Ardiansyah Saputra — Sat, 29 Jun 2024 00:00:00 +0000

The reviews given by visitors are critical and should be meticulously analyzed to enhance the quality of hotel services. Traditionally, reviewing each guest’s feedback manually is time-consuming and inefficient. Therefore, there is a need for an effective technique to aggregate and analyze large volumes of reviews. This research aims to provide a sentiment analysis of hotel visitor reviews using the Naïve Bayes algorithm. This study's findings are based on its comprehensive approach to automating sentiment analysis, which significantly reduces processing time and increases accuracy. For this purpose, we used Google Colaboratory to implement and evaluate the Naïve Bayes algorithm. The results reveal that the sentiment analysis model achieves varying levels of accuracy across different hotels: Hotel Arena with 0.74, K Hotel George with 0.94, Hotel Claridge Paris with 0.92, Suite Hotel 900 m zur Oper with 0.84, and Atlantis Hotel Vienna with 0.86. These findings underscore the potential of the Naïve Bayes algorithm for effectively capturing and analyzing customer sentiment, offering a valuable tool for hotel management to understand and improve guest satisfaction. This study is pioneering in its application of the Naïve Bayes algorithm in the context of sentiment analysis for multiple hotel brands, providing a scalable solution for the hospitality industry.

The Implementation of the K-Means Clustering Algorithm for Awarding Scholarships to Outstanding Students

Sat, 29 Jun 2024 00:00:00 +0000

Offering scholarships is an important tactic to promote higher education and scholastic success among students. The purpose of this study is to improve the impartiality and efficiency of scholarship awarding decision making by developing a decision support system built based on the K-Means clustering method. Students were grouped based on relevant variables and academic achievement requirements using the K-Means clustering algorithm. This approach creates homogeneous groupings based on scholarship recipients' history and academic performance data. The result of this clustering helps in recognizing trends and attributes that form the basis for future scholarship grant choices. This strategy was put into practice by creating a decision support system linked to student information and academic tracking. On processing the clustering data, cluster 0 with the status of eligible scholarship recipients amounted to 43 data and the cluster contained 157 data. It seems to contain data. Cluster 1 includes the status of scholarship recipients who do not meet the requirements. From the results of data analysis, it can be concluded that the scholarship recipient students are really outstanding students.

Sentiment Analysis of Tiktok Shop Closure using Naïve Bayes Algorithm and Support Vector Machine

Ahmad Zamsuri Ahmad, Wileks Wileks — Sat, 29 Jun 2024 00:00:00 +0000

This research explores the closure of TikTok Shops in Indonesia triggered by the implementation of Minister of Trade Regulation (MOT) 31/2020. It focuses on analyzing user responses and sentiment towards this policy by utilizing Naïve Bayes and Support Vector Machine algorithms. While some netizens supported the move for security reasons, while others criticized it for limiting business opportunities, an analysis of 1000 datasets from Twetter with the keyword "close TikTok Shop" revealed that neutral sentiment dominated, indicating a lack of clarity or confidence regarding the reason for the closure. The results also showed that Support Vector Machine (SVM) had higher accuracy (94.4%) than Naïve Bayes (89.1%), signaling the superiority of SVM in classifying sentiment in this dataset. These findings provide deep insights into the public's perceptions and attitudes regarding the closure of TikTok Shop, providing an important basis for government and corporate understanding of the public's response to the policy

Review of Machine Learning Algorithm for Intrusion Detection System

Guntoro Guntoro — Sat, 29 Jun 2024 00:00:00 +0000

Intrusion Detection Systems (IDS) are essential components in cybersecurity that aim to detect, identify, and mitigate threats to information systems. In recent years, the application of machine learning algorithms has significantly enhanced the effectiveness of IDS. This systematic literature review (SLR) analyzes and summarizes research studies on IDS using machine learning techniques from 2019 to 2023. The review focuses on key aspects such as datasets, machine learning algorithms, and types of attacks detected. The analysis reveals that Support Vector Machine (SVM) and Random Forest (RF) are the most frequently employed algorithms due to their high accuracy and robustness. Datasets such as NSL-KDD, KDD-Cup’99, and UNSW-NB15 are commonly used for training and evaluating IDS models. Various attack types, including Denial of Service (DoS), User to Root (U2R), Remote to Local (R2L), and Probing, are addressed in these studies. This SLR highlights the strengths and limitations of different machine learning approaches in IDS, offering insights into current trends and future research directions. The findings suggest a growing trend towards the use of ensemble methods and optimization techniques to improve IDS performance. Additionally, the review underscores the importance of diverse and realistic datasets for the accurate evaluation of IDS models. This comprehensive analysis aims to provide researchers and practitioners with a detailed understanding of the advancements in IDS using machine learning, guiding future research and development in this critical area of cybersecurity

Monitoring on Solid Foundation Servers Using Suricata Through Telegram Bot Notifications

Muhammad Faishal Gusrianda, Fajrizal Fajrizal, Guntoro Guntoro Guntoro — Mon, 01 Jul 2024 00:00:00 +0000

This research was conducted to study the Telegram Bot-based Suricata technique, which is implemented in the application of data center server security where the data center is related to the center of the Drug and Narcotics Agency of the Republic of Indonesia. We collected data from observations and interviews conducted in the administration section of server data security. Furthermore, the data is analyzed to study the pattern of each technique used. Based on this pattern, a better technique is chosen for the security process to be carried out. Based on the experiments conducted, it is clear that each of these techniques has advantages and disadvantages. However, this Telegram Bot-Based Suricata technique is more recommended because it is simpler in implementation, the desired results can be achieved, and the system can be monitored anywhere.