Digital Zone: Jurnal Teknologi Informasi dan Komunikasi

Sentiment Analysis on the Construction of the Jakarta-Bandung High-Speed Train on Twitter Social Media Using Recurrent Neural Networks Method

Titan Kinan Salaatsa, Yuliant Sibaroni — Sat, 29 Oct 2022 00:00:00 +0000

During the construction of the Jakarta-Bandung high-speed train, many Indonesian people gave their responses to the public. The answers were also varied, with some giving positive and negative reactions. The purpose of this study is to analyze the sentiments of the responses given by the public to the construction of the Jakarta-Bandung high-speed train on Indonesian-language Twitter. To perform sentiment analysis, tweet data was collected utilizing data crawling based on keywords related to the construction of the Jakarta-Bandung high-speed train and given positive, negative, and neutral labels and then represented into numbers using the Keras tokenizer. The method used for sentiment classification of tweet data is the Recurrent Neural Networks method. The highest accuracy results were obtained using the GRU architecture with an accuracy of 69.62%.

Live Forensics Analysis Of Malware Identified Email Crimes To Increase Evidence Of Cyber Crime

Yudhi Prawira Prawira, Samsudin Samsudin — Wed, 09 Nov 2022 00:00:00 +0000

Now days Email is the most important aplplication on the internet, this make email one of the industry’s most targeted sector for commiting cyber crimes. Email phishing and spam not only harm many parties but also consumes a lot of network bandwidth. Most spam are emotet malware. Trojan malware that targets internet users financial system to steal financial information and personal data by sending phishing. In this research, digital forensics analysis email crimes identified malware using live forensics and tools analyze digital evidence of email content, as wall as offVise, Wireshark, and Procmon to analyze malware activities. The results of the investigation of the email content carried out using software found digital evidence that could be used as a reference that attachment downloaded by the victim was Emoted type malware, when the victim opened it, this malware will be installed automatically on the victim’s computer.

Sentiment Analysis of Public Opinion Regarding Papuan Local Languages Condition Using Data Science Approach

Nur Fitrianingsih Hasan, Aisyah Aisyah, Rahman Rahman, Herlin Wonda — Sat, 26 Nov 2022 00:00:00 +0000

Regional languages can support economic empowerment and improvement through the tourism sector. Opinions from people's expressions in social media and online news collections in reporting the condition of regional languages often become headlines in cyberspace that number in the thousands, which can be used as new knowledge as a basis for making decisions through the mining method. This study aims to explore public opinion sentiment related to the condition of the Papuan language, sourced from text data in cyberspace using a data science approach, namely the classification method with text mining techniques using the naïve bayes algorithm. Public opinion sentiments are processed and the results are presented using word cloud visualization through 4 stages of data science, namely data collection, data preprocessing, modeling exploration and visualization analysis. The result of 778 opinions, 92% tend to have a positive sentiment. The analysis of public opinion sentiment is carried out by the naïve bayes algorithm which has an algorithm model accuracy of 78% and a precision of 88%. The machine learning model that was built and the word cloud visualization analysis succeeded in providing new insights regarding the condition of the Papuan language.

SVM Method with FastText Representation Feature for Classification of Twitter Sentiments Regarding the Covid-19 Vaccination Program

Mukti M Kusairi, Surya Agustian — Sat, 26 Nov 2022 00:00:00 +0000

Covid-19 is a virus that has a high level of spread, making the government implement a mass vaccination program throughout Indonesia. This program received a lot of responses from the public, with positive and negative opinions or comments. Currently, the public's response through social media is also an input and consideration for the government to implement a program. Therefore, this study was conducted to produce a method approach to assessing the Covid-19 vaccination program by calculating the percentage of each sentiment class. The method used is the Support Vector Machine (SVM) and the fasttext language model feature as a representation of words in the Covid-19 vaccination sentiment dataset collected from Twitter. The data used has been dataset balancing, feature selection and parameter tuning, the optimal SVM model is obtained with a composition of 2536 training data, 778 development data and testing of 400 testing data, resulting in the best value of fi-1 score of 59% with an accuracy rate of 68%. The system is quite successful in detecting sentiment in tweets compared to before.
Keywords: sentiment classification, FastText, SVM, Covid-19 vaccine.

C4.5 Algorithm Implementation For Public Sentyment Analysis Covid-19 Vaccine

Devi Astri Nawangnugraeni, M. Zakki Abdillah, Akrim Teguh Suseno — Wed, 30 Nov 2022 00:00:00 +0000

Corona virus disease is one of the dangerous diseases and has been prevented by giving vaccinations. In an effort to prevent, there must be a positive or negative public response. One of the media facilities used to convey public responses is Twitter. The public's reaction can be analyzed using public sentiment analysis using C4.5 algorithm. The purpose of paper for determine public's response to the administration of moderna and pfizer vaccinations. The implemented methodology starts from collecting data taken from tweets, pre-processing, classification using the C4.5 algorithm and validation using k-fold cross validation. Based on the results of the moderna keyword analysis, the positive sentiment response was 6% and negative sentiment was 94%, while the pfizer keyword positive sentiment was 12.4% and negative sentiment was 87.6%. The results of test iteration that have been carried out 3 times, the average error value is 38%.

Speech Recognition for English Sentences with Malay Accent

Keumala Anggraini, Lucky Lhaura Van FC, Yuvi Darmayunata — Wed, 30 Nov 2022 00:00:00 +0000

Some countries conduct speech research using accents. One language that has a different accent is English. English is an international language that is often used to communicate with citizens of other countries. For beginner, many difficulty to translate English with accent, include malay accent. This study performs speech recognition using English and Riau Malay accents. This research use Google Recognizer to cut words in sentences, Mel Frequency Cepstral Coefficient for feature extraction, and Hidden Markov Model for classification. The accuracy of this research is 94.02%.