Text Mining Classification Research Articles

The Accuracy Improvement of Text Mining Classification on Hospital Review through The Alteration in The Preprocessing Stage

Sentiment analysis is a part of text mining used to dig up information from a sentence or document. This study focuses on text classification for the purpose of a sentiment analysis on hospital review by customers through criticism and suggestion on Google Maps Review. The data of texts collected still contain a lot of nonstandard words. These nonstandard words cause problem in the preprocessing stage. Thus, the selection and combination of techniques in the preprocessing stage emerge as something crucial for the accuracy improvement in the computation of machine learning. However, not all of the techniques in the preprocessing stage can contribute to improve the accuracy on classification machine. The objective of this study is to improve the accuracy of classification model on hospital review by customers for a sentiment analysis modeling. Through the implementation of the preprocessing technique combination, it can produce a highly accurate classification model. This study experimented with several preprocessing techniques: (1) tokenization, (2) case folding, (3) stop words removal, (4) stemming, and (5) removing punctuation and number. The experiment was done by adding the preprocessing methods: (1) spelling correction and (2) Slang. The result shows that spelling correction and Slang method can assist for improving the accuracy value. Furthermore, the selection of suitable preprocessing technique combination can fasten the training process to produce the more ideal text classification model.

Comparison of Text Mining Classification Algorithms in Interbank Money Transfer Application

Funds transfer is a series of orders from the sender whose purpose is to move money from the sender to the recipient. The high interbank transaction fees imposed on each bank makes people use an interbank money transfer application, interbank money transfer transactions such as the Flip application are much in demand by the public because there are no administrative fees imposed on users. Opinion of the users of the application is processed using a text mining classification algorithm, namely the Naïve Bayes Algorithm and k-NN, the two algorithms are compared to produce which algorithm has high accuracy in processing the opinion of the flip money transfer application. Based on this matter, researchers conducted a sentiment analysis of the Flip Application, K-Nearest Neighbor (k-NN). After conducting research on sentiment analysis of Flip Applications, the Naïve Bayes classification algorithm has an accuracy of 91.25% and an ROC curve with an AUC value of 0.500. Whereas K-Nearest Neighbor has an accuracy of 85.25% and an ROC curve with an AUC value of 0.937. The Naïve Bayes algorithm can be said to be ”good classification” and the public can make the decision to use the Flip Application.

CLASSIFICATION OF CUSTOMER COMPLAINTS ON INSTAGRAM COMMENTS USING NAÏVE BAYES ALGORITHM WITH N-GRAM FEATURE EXTENSION

Customer complaints about the company can be used as a form of self-evaluation and performance that has been carried out by the company, based on customer complaints the company can find out the weaknesses that exist in the company and fix them. The forms of submitting customer complaints are very diverse, currently not only by telephone, but customers also submit suggestions or complaints, customers can submit suggestions or complaints via electronic mail or e-mail or forums in cyberspace that are indeed created by product-producing companies to accommodate various complaints, suggestions, and direct criticism from consumers, especially social media that are free to express opinions on the delivery services used. Instagram is a social media that is more inclined towards images and on the other hand, has captions and comments text, a study is needed for the problem of customer complaints from shipping service users on an Instagram account of a delivery service company. Based on this background, a solution is needed in solving problems for text mining classification using Naïve Bayes with SMOTE techniques and N-Gram feature extraction with the usual process for text mining so that it can produce Naïve Bayes and SMOTE accuracy with an accuracy of 88.54%, before implementation. N-Gram and the accuracy rate increased by 1.44% after the N-Gram Term was applied to 89.98% by using a dataset of 776 Instagram comment text records that had to preprocess text.

Text Mining In Healthcare

In healthcare, data mining intensively and extensively becoming essential. Data mining applications can benefit all patients and the healthcare professionals. This paper starts with introducing data mining and the healthcare paradigm. This study confers various techniques of data mining in healthcare application domain. As the scope of the study is limited to text mining classification, state of art in particular to healthcare text mining classification is studied in detail with suggested improvements. Various issues and challenges owing to the type of data in healthcare are also discussed in detail with possible solutions. Finally, the paper highlights the need for personalized prescriptive systems for patients and healthcare professionals

Implementation of Text Mining Classification as a Model in the Conclusion of Tafsir Bil Ma'tsur and Bil RA'YI Contents

Studies related to Tafsir Qur'an have only been carried out based on a manual system or on application-based development. The purpose of this study is to build an application which can classify type of interpretation automatically into two classes, tafsir Bil Ma'tsur and tafsir Bil Ra'yi, can provide user convenience in the Al-Qur-an. KNN algorithm is a reliable algorithm in the classification process, also has many parts of the algorithm. This study was done by applying K-Nearest Neighbor (KNN) algorithm with an accuracy of 98.12%. However KNN had been compared firstly with Modified K-Nearest Neighbor (MKNN) and Fuzzy K-Nearest Neighbor (FKNN) algorithms, where the two algorithms had 98.01% and 88.3% accuracy respectively. MKNN is the best algorithm with the highest accuracy, but also has a high error value by 4.3% which is higher than KNN, 1.9%. From the research conducted, the more text documents used in KNN modeling, the higher accuracy will be. Therefore, in the implementation of this application KNN is used as modeling in the conclusion of Tafsir Al-Qur'an. Tests performed with BlackBox Testing and User Aceptance Test reached value of 100% and 98.8%.

Arabic Poetry Authorship Attribution using Machine Learning Techniques

In this study, authorship attribution in Arabic poetry will be conducted to determine the authorship of a specified text after documents with recognized authorships have been allocated. This work also measures the impact performance of Naïve Bayes, Support Vector Machine and Linear discriminant analysis for Arabic poetry authorship attribution using text mining classification. Several features such as lexical features, character features, structural features, poetry features, syntactic features, semantic features and specific word features are utilized as the input data for text mining, using classification algorithms Linear discriminant analysis, Support Vector Machine and Naïve Bayes by Arabic Poetry Authorship Attribution Model (APAAM). The dataset of Arabic poetry is divided into two sets: known poetic in training dataset texts and anonymous poetic texts in a test dataset part. In the experiment, a set of 114 random poets from entirely different eras are used. The highest performance accuracy value is 99, 12%; the performance rate at the attribute level is 98.246%; the level of techniques is 92.836%.

Comparative Analysis of Text Mining Classification Algorithms for English and Indonesian Qur’an Translation

Qur'an, As-Sunnah, and Islamic old book have become the sources for Islam followers as sources of knowledge, wisdom, and law. But in daily life, there are still many Muslims who do not understand the meaning of the sentence in the Qur'an even though they read it every day. It becomes a challenge for Science and Engineering field academicians especially Informatics to explore and represent knowledge through intelligent system computing to answer various questions based on knowledge from the Qur'an. This research is creating an enabling computational environment for text mining the Qur'an, of which purpose is to facilitate people to understand each verse in the Qur'an. The classification experiment uses Support Vector Machine (SVM), Naive Bayes, k-Nearest Neighbor (kNN), and J48 Decision Tree classifier algorithms with Al-Baqarah verses translated to English and Indonesian as the dataset which was labeled by three most fundamental aspects of Islam: 'Iman' (faith), 'Ibadah' (worship), and 'Akhlaq' (virtues). Indonesian translation was processed by using the sastrawi package in Python to do the pre-processing and StringToWord Vector in WEKA with the TF-IDF method to implement the algorithms. The classification experiments are determined to measure accuracy, and f-measure, it tested with a percentage split 66% as the data training and the rest as the data testing. The decision from an experiment that was carried out by the classification results, SVM classifier algorithms have the overall best accuracy performance for the Indonesian translation of 81.443% and the Naïve Bayes classifier has the best accuracy for the English translation, which achieved 78.35%.

Document Categorization Using Decision Tree: Preliminary Study

This preliminaries study aims to propose a good classification technique that capable of doing document classification based on text mining technique and create an algorithm to automatically classify document according to its folder based on document’s content while able to do sentiment analyses to data sets and summarize it. The objective of this paper to identify an efficient text mining classification technique which can resulted with highest accuracy of classifying document into document folder, capable of extracting valuable information from context-based term that can be used as an output for algorithm to do automatic classification and evaluate the classification technique. Methodology of this study comprises in 5 modules which is 1) Document collection, 2) Pre-Processing Stage, 3) Term Frequency-Inversed Document Frequency, 4) Classification Technique and Algorithm, and lastly 5) Evaluation and Visualization of the classification result. The proposed framework will have utilized Term Frequency-Inversed Document Frequency (TF-IDF) and Decision Tree technique which TF-IDF used as purposes to rank all the terms based on most frequent to least frequent terms so, while decision tree function as decision making in terms of deciding which folder the document belongs to.

What is wrong with topic modeling? And how to fix it using search-based software engineering

ContextTopic modeling finds human-readable structures in unstructured textual data. A widely used topic modeling technique is Latent Dirichlet allocation. When running on different datasets, LDA suffers from “order effects”, i.e., different topics are generated if the order of training data is shuffled. Such order effects introduce a systematic error for any study. This error can relate to misleading results; specifically, inaccurate topic descriptions and a reduction in the efficacy of text mining classification results. ObjectiveTo provide a method in which distributions generated by LDA are more stable and can be used for further analysis. MethodWe use LDADE, a search-based software engineering tool which uses Differential Evolution (DE) to tune the LDA’s parameters. LDADE is evaluated on data from a programmer information exchange site (Stackoverflow), title and abstract text of thousands of Software Engineering (SE) papers, and software defect reports from NASA. Results were collected across different implementations of LDA (Python+Scikit-Learn, Scala+Spark) across Linux platform and for different kinds of LDAs (VEM, Gibbs sampling). Results were scored via topic stability and text mining classification accuracy. ResultsIn all treatments: (i) standard LDA exhibits very large topic instability; (ii) LDADE’s tunings dramatically reduce cluster instability; (iii) LDADE also leads to improved performances for supervised as well as unsupervised learning. ConclusionDue to topic instability, using standard LDA with its “off-the-shelf” settings should now be depreciated. Also, in future, we should require SE papers that use LDA to test and (if needed) mitigate LDA topic instability. Finally, LDADE is a candidate technology for effectively and efficiently reducing that instability.

Construction accident narrative classification: An evaluation of text mining techniques

Learning from past accidents is fundamental to accident prevention. Thus, accident and near miss reporting are encouraged by organizations and regulators. However, for organizations managing large safety databases, the time taken to accurately classify accident and near miss narratives will be very significant. This study aims to evaluate the utility of various text mining classification techniques in classifying 1000 publicly available construction accident narratives obtained from the US OSHA website. The study evaluated six machine learning algorithms, including support vector machine (SVM), linear regression (LR), random forest (RF), k-nearest neighbor (KNN), decision tree (DT) and Naive Bayes (NB), and found that SVM produced the best performance in classifying the test set of 251 cases. Further experimentation with tokenization of the processed text and non-linear SVM were also conducted. In addition, a grid search was conducted on the hyperparameters of the SVM models. It was found that the best performing classifiers were linear SVM with unigram tokenization and radial basis function (RBF) SVM with uni-gram tokenization. In view of its relative simplicity, the linear SVM is recommended. Across the 11 labels of accident causes or types, the precision of the linear SVM ranged from 0.5 to 1, recall ranged from 0.36 to 0.9 and F1 score was between 0.45 and 0.92. The reasons for misclassification were discussed and suggestions on ways to improve the performance were provided.

String Kernels for Native Language Identification: Insights from Behind the Curtains

The most common approach in text mining classification tasks is to rely on features like words, part-of-speech tags, stems, or some other high-level linguistic features. Recently, an approach that uses only character p-grams as features has been proposed for the task of native language identification (NLI). The approach obtained state-of-the-art results by combining several string kernels using multiple kernel learning. Despite the fact that the approach based on string kernels performs so well, several questions about this method remain unanswered. First, it is not clear why such a simple approach can compete with far more complex approaches that take words, lemmas, syntactic information, or even semantics into account. Second, although the approach is designed to be language independent, all experiments to date have been on English. This work is an extensive study that aims to systematically present the string kernel approach and to clarify the open questions mentioned above.A broad set of native language identification experiments were conducted to compare the string kernels approach with other state-of-the-art methods. The empirical results obtained in all of the experiments conducted in this work indicate that the proposed approach achieves state-of-the-art performance in NLI, reaching an accuracy that is 1.7% above the top scoring system of the 2013 NLI Shared Task. Furthermore, the results obtained on both the Arabic and the Norwegian corpora demonstrate that the proposed approach is language independent. In the Arabic native language identification task, string kernels show an increase of more than 17% over the best accuracy reported so far. The results of string kernels on Norwegian native language identification are also significantly better than the state-of-the-art approach. In addition, in a cross-corpus experiment, the proposed approach shows that it can also be topic independent, improving the state-of-the-art system by 32.3%.To gain additional insights about the string kernels approach, the features selected by the classifier as being more discriminating are analyzed in this work. The analysis also offers information about localized language transfer effects, since the features used by the proposed model are p-grams of various lengths. The features captured by the model typically include stems, function words, and word prefixes and suffixes, which have the potential to generalize over purely word-based features. By analyzing the discriminating features, this article offers insights into two kinds of language transfer effects, namely, word choice (lexical transfer) and morphological differences. The goal of the current study is to give a full view of the string kernels approach and shed some light on why this approach works so well.