Sriwijaya Journal of Informatics and Applications (SJIA)
Not a member yet
49 research outputs found
Sort by
Cross-Site Scripting Attack Detection using Rule-Based Signature
Get PDFRule-Based Signature or also known as Misuse Detection is IDS which rely on matching data captured on retrieval of attack pattern which in system that allow attacks. If the attack activity detected according to existing signature, then it will be read by system and called as attack. The advantage of this Signature-Based IDS is the accuracy of detecting matched attack which in the system with low false-positive result and high true-positive. Cross-Site Scripting is type of attack which is perform by injecting code (usually) JavaScript to a site. XSS is very often utilized by attacker to steal web browser resource such as cookie, credentials, etc. Dataset which used in this research is dataset which created by injecting script into a website. Once obtained the dataset, then feature extraction is performed to separate the attribute which used. XSS attack pattern can be easily recognized from URI, and then detected using engine which has been created. Detection result of algorithm which used is evaluated using confusion matrix to determine detection accuracy value which performed. Obtained accuracy detection of research result reached 99.4% with TPR 98.8% and FPR 0%
The Effect of Brill Tagger on The Classification Results of Sentiment Analysis Using Multinomial Naïve Bayes Algorithm
Get PDFTwitter is a good indicator for influence in research, the problem thatarises in research in the field of sentiment analysis is the large numberof factors such as the use of informal or colloquial language and otherfactors that can affect the results of sentiment classification. Toimprove the results of sentiment classification, an informationextraction process can be carried out. One part of the informationextraction feature is a part of speech tagging, which is the giving ofword classes automatically. The results of part of speech tagging areused for weighting words based on part of speech. This studyexamines the effect of Part of Speech Tagging with the method BrillTagger in sentiment analysis using the Naive Bayes Multinomialalgorithm. Testing were carried out on 500 twitter tweet texts andobtained the results of the sentiment classification with implementingpart of speech tagging precision by 73,2%, recall by 63,2%, f-measureby 67,6%, accuracy by 60,7% and without implementing part ofspeech tagging precision by 65,2%, recall by 60,6%, f-measure by62,4% accuracy by 53,3%. From the results of the accuracy obtained,it shows that the application of part of speech tagging in sentimentanalysis using the Multinomial Naïve Bayes algorithm has an effectwith an increase in classification performance
Text Summarization with K-Means Method
Get PDFText Summarization is a tool used to generate a short form of text that contains important information that is needed by the user automatically. In this study, Text Summarization was conducted on Indonesian news using K-Means method. The news is taken from CNN Indonesia with a free topic. K-Means is used to classify sentences that already have weight in the news with 2 clusters, namely text summaries and not text summaries. The initial centroid is selected based on the sentence with the largest value and the sentence with the smallest value. The test conducted on Indonesian news with a total 50 news and tested for feasibility using a questionnaire. K-Means was successfully summarizing the news with an average 27.3 % of original news length and gain 87% good summarize based on respondents from questionnaire
Fuzzy Time Series Optimization using Particle Swarm Optimization for Forecasting the Number of Fresh Fruit Bunches (FBB) of Palm Oil
Get PDFPalm oil is a reliable vegetable oil producer because the oil produced has advantages than oils from other plants. The amount of Fresh Fruit Bunches (FFB) raw material from Palm oil has a significant impact on the palm oil production process. Therefore, we need a method to forecasting the amount of palm oil (FFB). One of the suitable forecasting methods is fuzzy time series (FTS). However, FTS still has shortcomings such as innacurate determination of the interval length. For this reason, we need to optimize FTS interval to get optimal forecasting. This research implements Particle Swarm Optimization as the optimization method, FTS Chen-Hsu as the forecast method, and Mean Absolute Percentage Error (MAPE) as the measurement of error. The optimization result using PSO produce an error value of 2.0262% smaller than FTS 3.7108%
Face Detection Using Randomized Hough Transform (RHT) with Various Ellipses Segmentations
Get PDFFace detection is one of earlier phase in face recognition process. This research aims to get the faces area on digital image without being affected by face orientation, lights condition, background and the expression. The detected face area is usually shaped by a rectangle. Many pixels on the rectangle are not part of face, especially at the four of the image corners. This research use an ellipse as replacement a rectangle. The detected face is shaped by ellipses with various sizes and orientations. The digital image segmentations is used to detect face candidates area. The ellipse is formed by using Randomized Hough Transform (RHT) method, which is influenced by the center point of ellipse candidates. RHT found three random pixels on segmented image. The rate of success of RHT is determined by segmentation results. The research result is tested by using various thresholds, and get the best accuracy at 74.4%. The rate of accuracy is measured by comparing between RHT ellipses shape and circle shape on OpenCV library as ground truth
Spelling Checker using Algorithm Damerau Levenshtein Distance and Cosine Similarity
Get PDFWriting is an embodiment of the author's ideas that are to be conveyed to others. A writer often experiences typos in typing the script, so that it can influence the meaning of the text. Therefore, a system is needed to detect word errors. In this study, checking is done by using the Dictionary Lookup method and giving the candidate words using the Damerau Levenshtein Distance algorithm. Candidates will then determine the ranking by breaking the word into Bigram form and calculating the similarity value using the Cosine Similarity algorithm. The test results based on the data used yield different Mean Reciprocal Rank (MRR) values for each type of error. The type of error deletion produces an MRR value of 88.89%, the type of insertion error produces an MRR value of 97.78%, the type of substitution error produces an MRR value of 88.89%, the type of transposition error produces an MRR value of 89
Effect of N-Gram on Document Classification on the Naïve Bayes Classifier Algorithm
Get PDFNews has become a major need for everyone, with news we can get the information needed. News can be distributed in the form of print mass media, electronic mass media and online media. The means of spreading the news now have grown very rapidly, making the amount of information being managed are bigger and word management classified also not small. herefore, we need a system for classifying documents that are not structured. In this study, word processing in a document is done by N-Gram as a feature generation. The document classification process is carried out using the Naïve Bayes Classifier algorithm. This study examines the effect of N-Gram on document classification on the Naïve Bayes Classifier algorithm. The results of the classification accuracy of documents by applying N-Gram is 32.68% and without applying N-Gram is 84.97%. A decrease in the classification results occurs the number of features that result from solving N-Gram that is unique or dominant to another category. The accuracy of the results obtained shows that the application of N-Gram in the classification of documents using the Naïve Bayes Classifier algorithm gives a decreased effect on the performance of the classificatio
Urea Fertilizer Quality Testing with Chi-Squared Automatic Interaction Detection (CHAID) Algorithm
Get PDFPT. XYZ has a Laboratory section in each of its factories that performs its duties manually to determine the quality of the fertilizers to be produced. This manual method is most likely at risk of human error and causes errors in the results of determining the quality of urea fertilizer. An expert system was built using the Chi-Squared Automatic Interaction Detection (CHAID) algorithm which can test the quality of urea fertilizer. The CHAID algorithm applies the decision tree technique where the technique will always branch off two or more as a basis in establishing rules. The system takes the values of the urea fertilizer test parameters as attributes. These attributes are processed to produce the most significant values that will be branches in the decision tree. The parameters used include Nitrogen, Biuret, Moisture, Free Ammonia, Iron, Oil Content, Crushing Strength, and Size Distribution. CHAID algorithm is suitable to be used to test the quality of urea fertilizer because in this study produced 4 different decision trees with an accuracy value of 99% using as much as 100 test data. This number influenced by the amount of training data used to build the rules
Multilabel Classification for News Article Using Long Short-Term Memory
Get PDFMultilabel text classification is a task of categorizing text into one or more categories. Like other machine learning, multilabel classification performance is limited when there is small labeled data and leads to the difficulty of capturing semantic relationships. In this case, it requires a multi-label text classification technique that can group four labels from news articles. Deep Learning is a proposed method for solving problems in multi-label text classification techniques. By comparing the seven proposed Long Short-Term Memory (LSTM) models with large-scale datasets by dividing 4 LSTM models with 1 layer, 2 layer and 3-layer LSTM and Bidirectional LSTM to show that LSTM can achieve good performance in multi-label text classification. The results show that the evaluation of the performance of the 2-layer LSTM model in the training process obtained an accuracy of 96 with the highest testing accuracy of all models at 94.3. The performance results for model 3 with 1-layer LSTM obtained the average value of precision, recall, and f1-score equal to the 94 training process accuracy. This states that model 3 with 1-layer LSTM both training and testing process is better. The comparison among seven proposed LSTM models shows that model 3 with 1 layer LSTM is the best model