Classification Of Textual Data Research Articles

ВІД АНАЛІЗУ ТЕКСТУ ДО МОДЕЛЮВАННЯ ПРИРОДНОЇ МОВИ: КОМПЛЕКСНЕ ДОСЛІДЖЕННЯ

The article presents a comprehensive analysis of modern techniques and technologies used to analyze the tonality of textual data. Through a series of experiments, basic machine and deep learning algorithms such as sentiment analysis were developed to automatically detect and classify emotional tones in texts. Practical applications of these methods are also explored, ranging from social network monitoring to expert analysis and analysis of user reviews. Next, the article delves into the classification of textual data using modern machine and deep learning methods. This section of the paper reviews several classification models, including the Bayesian most-narrow classifier, support vector methods, and neural networks, highlighting their advantages and limitations. In addition, it emphasizes the importance of implementing text classification in various fields, including social network analysis, news article categorization, and automated document processing. In addition, clustering of similar textual data is considered for further analysis. Various clustering algorithms, such as k-means, hierarchical clustering, and spectral clustering, are compared, with special emphasis on their application to large text corpora. A demonstration of the practical application of text clustering, including data organization, topic search, and style identification in paper works, was also demonstrated. Moving on, the article considers the processing of individual thematic structures in textual data and its further analysis. An in-depth analysis of topic modeling techniques, such as the realized Dirichlet distribution (LDA) model, as well as its capabilities and limitations, is explored. Practical applications of topic modeling are demonstrated in various ways, including text collection analysis, news trend detection, and automatic document categorization. Finally, the article discusses some challenges and future prospects for the development of thematic modeling.

FFO-ABC DepressioGuard: A Hybrid Classification Framework for Social Media Depression Detection

In the phase of digital communication, social media sites have developed into a major centre for individuals to express their ideas, emotions, and views. Amidst this influx of user-generated content, mental health conditions, notably depression, have garnered increasing attention due to their pervasive impact on individuals and societies. Early detection and intervention are crucial in managing and preventing its adverse effects. As a consequence an innovative machine learning (ML) textual data classification framework is designed to detect depression through social media streams, employing a Firefly-Optimized Support Vector Machine (FFO-SVM) and Artificial Bee Colony (ABC) Optimized SVM classifiers. Initially, data collection and preprocessing are performed, followed by feature extraction using Time Frequency-Inverse Document Frequency (TF-IDF). After extraction of features classification is performed using FFO-SVM and ABC-SVM classifiers. To tune the parameters of SVM to work efficiently, FFO and ABC are employed. The proposed framework combines the power of ML with the optimization capabilities of the ABC and FFO Algorithm to enhance classification accuracy. Through extensive experimentation and analysis, the framework's performance is evaluated using relevant metrics. Results indicated that proposed classification techniques outperformed conventional methods, showcasing its effectiveness in handling the complexity of depression detection from social media data.

Multi-Label Annotation and Classification of Arabic Texts Based on Extracted Seed Keyphrases and Bi-Gram Alphabet Feed Forward Neural Networks Model

In natural language processing, text classification is a fundamental problem. Multi-label classification of textual data is a challenging topic in text classification where an instance can be associated with more than one label. This paper presents a multi-label annotation and classification methodology for Arabic text data that is not currently classified as multi-label, aiming to analyze and compare the performance of various multi-label learning approaches. The current work includes two phases: The first involves automatic annotation of hotel reviews with more than one label based on the aspects found in the reviews. In this phase, review data instances were automatically annotated as multi-label based on the extracted seed keyphrases clusters. The second phase involves experiments to compare the performance of various multi-label classification learning methods. In this phase, we introduced different models including a feed-forward networks model that learns a vector representation based on the bi-gram alphabet rather than the commonly used bag-of-words model. The bi-gram alphabet vector representation model has the advantage of having reduced feature dimensions and not requiring natural language processing tools. The results indicated that employing the bi-gram alphabet vector representation feed forward neural network is a competitive solution for the multi-label text classification problem. It has achieved an accuracy of about 75.2%, and standard deviation (0.062).

An efficient approach for textual data classification using deep learning.

Text categorization is an effective activity that can be accomplished using a variety of classification algorithms. In machine learning, the classifier is built by learning the features of categories from a set of preset training data. Similarly, deep learning offers enormous benefits for text classification since they execute highly accurately with lower-level engineering and processing. This paper employs machine and deep learning techniques to classify textual data. Textual data contains much useless information that must be pre-processed. We clean the data, impute missing values, and eliminate the repeated columns. Next, we employ machine learning algorithms: logistic regression, random forest, K-nearest neighbors (KNN), and deep learning algorithms: long short-term memory (LSTM), artificial neural network (ANN), and gated recurrent unit (GRU) for classification. Results reveal that LSTM achieves 92% accuracy outperforming all other model and baseline studies.

An Improved Arabic Sentiment Analysis Approach using Optimized Multinomial Naïve Bayes Classifier

Arabic sentiment analysis has emerged during the last decade as a computational process on Arabic texts for extracting people's attitudes toward targeted objects or their feelings and emotions regarding targeted events. Sentiment analysis (SA) using machine learning (ML) methods has become an important research task for developing various text-based applications. Among different ML classifiers, multinomial Naïve Bayes (MNNB) classifier is widely used for documents classification due to its ability for performing statistical analysis of text contents. It significantly simplifies textual-data classification and offers an alternative to heavy ML-based semantic analysis methods. However, the MNNB classifier has a number of hyper-parameters affects the classification task of texts and controls the decision boundary of the model itself. In this paper, an optimized MNNB classifier-based approach is proposed for improving Arabic sentiment analysis. A number of experiments are conducted on large sets of Arabic tweets to evaluate the proposed approach. The optimized MNNB classifier is trained on three datasets and tested on a different separated test set to show the performance of developed approach. The experimental results on the test set revealed that the optimized MNNB classifier of proposed approach outperforms the traditional MNNB classifier and other baseline classifiers. The accuracy rate of the optimization approach is increased by 1.6% compared with using the default values of the classifier’s hyper-parameters.

Application of machine learning in the process of classification of advertised jobs

Institutions that provide official statistics tend to use external data sources such as administrative data sources besides regular statistical surveys. In addition to the mentioned data sources, Big Data became recognized as a new data source for the provider of official statistics. Classification of textual data is one of the elementary tasks for the provider of official statistics, regardless of data sources. In this paper, application of traditional machine learning algorithms, Multinomial Naive Bayes and Support Vector Machine, for the classification of advertised jobs according to ISCO-08, has been presented. The paper presents the methods of collecting data on advertised jobs from four websites and procedures for creating a multilingual dataset. There are different types of text preprocessing, such as converting uppercase letters into lowercase letters, stopword removal, punctuation mark removal, lemmatization, correction of commonly misspelled words, and reduction of replicated characters. We hypothesized that the application of different combinations of preprocessing methods influenced the text classification results. Two experiments had conducted to test the hypothesis. Both experiments results showed that using the Support Vector Machine algorithm on a created dataset gives better results than Multinomial Naive Bayes. Performed experiments showed that the proposed algorithms gave a good performance with an overall accuracy of up to 90% but with different accuracy for individual classes due to an imbalanced dataset.

Post-hoc explanation of black-box classifiers using confident itemsets

Black-box Artificial Intelligence (AI) methods, e.g. deep neural networks, have been widely utilized to build predictive models that can extract complex relationships in a dataset and make predictions for new unseen data records. However, it is difficult to trust decisions made by such methods since their inner working and decision logic is hidden from the user. Explainable Artificial Intelligence (XAI) refers to systems that try to explain how a black-box AI model produces its outcomes. Post-hoc XAI methods approximate the behavior of a black-box by extracting relationships between feature values and the predictions. Perturbation-based and decision set methods are among commonly used post-hoc XAI systems. The former explanators rely on random perturbations of data records to build local or global linear models that explain individual predictions or the whole model. The latter explanators use those feature values that appear more frequently to construct a set of decision rules that produces the same outcomes as the target black-box. However, these two classes of XAI methods have some limitations. Random perturbations do not take into account the distribution of feature values in different subspaces, leading to misleading approximations. Decision sets only pay attention to frequent feature values and miss many important correlations between features and class labels that appear less frequently but accurately represent decision boundaries of the model. In this paper, we address the above challenges by proposing an explanation method named Confident Itemsets Explanation (CIE). We introduce confident itemsets, a set of feature values that are highly correlated to a specific class label. CIE utilizes confident itemsets to discretize the whole decision space of a model to smaller subspaces. Extracting important correlations between the features and the outcomes of the classifier in different subspaces, CIE produces instance-wise and class-wise explanations that accurately approximate the behavior of the target black-box. Conducting a set of experiments on various black-box classifiers, and different tabular and textual data classification tasks, we show that our CIE method performs better than the previous perturbation-based and rule-based explanators in terms of the descriptive accuracy (an improvement of 9.3%) and interpretability (an improvement of 8.8%) of the explanations. Subjective evaluations demonstrate that the users find the explanations of CIE more understandable and interpretable than those of the other comparison methods.

Comparison of Supervised Classification Models on Textual Data

Text classification is an essential aspect in many applications, such as spam detection and sentiment analysis. With the growing number of textual documents and datasets generated through social media and news articles, an increasing number of machine learning methods are required for accurate textual classification. For this paper, a comprehensive evaluation of the performance of multiple supervised learning models, such as logistic regression (LR), decision trees (DT), support vector machine (SVM), AdaBoost (AB), random forest (RF), multinomial naive Bayes (NB), multilayer perceptrons (MLP), and gradient boosting (GB), was conducted to assess the efficiency and robustness, as well as limitations, of these models on the classification of textual data. SVM, LR, and MLP had better performance in general, with SVM being the best, while DT and AB had much lower accuracies amongst all the tested models. Further exploration on the use of different SVM kernels was performed, demonstrating the advantage of using linear kernels over polynomial, sigmoid, and radial basis function kernels for text classification. The effects of removing stop words on model performance was also investigated; DT performed better with stop words removed, while all other models were relatively unaffected by the presence or absence of stop words.

Support vector machine parameter tuning based on particle swarm optimization metaheuristic

This paper introduces a method for linear support vector machine parameter tuning based on particle swarm optimization metaheuristic, which is used to find the best cost (penalty) parameter for a linear support vector machine to increase textual data classification accuracy. Additionally, majority voting based ensembling is applied to increase the efficiency of the proposed method. The results were compared with results from our previous research and other authors’ works. They indicate that the proposed method can improve classification performance for a sentiment recognition task.

Data Extraction and Sentimental Analysis from “Twitter” using Web Scrapping

In this paper , we attempt to do the sentimental analysis of the 2016 US presidential elections. Sentimental analysis requires the data to be extracted from websites or sources where people present their opinions, views ,complaints about the subjects that need to analyzed .Furthermore, it is necessary to ensure that the sample size of the data is large enough to get conclusive results .It is also necessary to ensure that the data is cleaned before it is used to make predictions. Cleaning is done using common techniques like tokenization, spell check ,etc. Sentimental Analysis is one of the by-products of Natural Language Processing . This paper includes data collection as well as classification of textual data based on machine learning.

ML-SLSTSVM: a new structural least square twin support vector machine for multi-label learning

Multi-label learning (MLL) is a special supervised learning task, where any single instance possibly belongs to several classes simultaneously. Nowadays, MLL methods are increasingly required by modern applications, such as protein function classification, speech recognition and textual data classification. In this paper, a structural least square twin support vector machine (SLSTSVM) classifier for multi-label learning is presented. This proposed ML-SLSTSVM focuses on the cluster-based structural information of the corresponding class in each optimization problem, which is vital for designing a good classifier in different real-world problems. This method is extended to a nonlinear version by the kernel trick. Experimental results demonstrate that proposed method is superior in generalization performance to other classifiers.

A New Synthetic Oversampling Method Using Ontology and Feature Selection in Order to Improve Imbalanced Textual Data Classification in Persian Texts

Ever-growing extension of textual data has increased the necessity of processing textual data. Data imbalance in classification of textual data is one of the cases that decrease efficiency. In order to confront with imbalance problem, various methods are suggested. Some of the methods are: data-based, cost-based, algorithm-based and feature selection methods. In recent researches, some methods are considered into account using ensemble methods. In this research, a new oversampling method is suggested. In the new method the number of minor class samples is increased using ontology and then random oversampling is performed for minor class. Finally, using the methods of feature selection, appropriate features are selected. New ensemble method was tested using Hamshahri data. The results show that the ensemble method on Hamshahri collection, despite decreasing number of features, causes the improvement of classification results for polynomial Naïve Bayes and decision tree.