TF-IDF Feature Research Articles

Automating fake news detection system using multi-level voting model

The issues of online fake news have attained an increasing eminence in the diffusion of shaping news stories online. Misleading or unreliable information in the form of videos, posts, articles, URLs is extensively disseminated through popular social media platforms such as Facebook and Twitter. As a result, editors and journalists are in need of new tools that can help them to pace up the verification process for the content that has been originated from social media. Motivated by the need for automated detection of fake news, the goal is to find out which classification model identifies phony features accurately using three feature extraction techniques, Term Frequency–Inverse Document Frequency (TF–IDF), Count-Vectorizer (CV) and Hashing-Vectorizer (HV). Also, in this paper, a novel multi-level voting ensemble model is proposed. The proposed system has been tested on three datasets using twelve classifiers. These ML classifiers are combined based on their false prediction ratio. It has been observed that the Passive Aggressive, Logistic Regression and Linear Support Vector Classifier (LinearSVC) individually perform best using TF-IDF, CV and HV feature extraction approaches, respectively, based on their performance metrics, whereas the proposed model outperforms the Passive Aggressive model by 0.8%, Logistic Regression model by 1.3%, LinearSVC model by 0.4% using TF-IDF, CV and HV, respectively. The proposed system can also be used to predict the fake content (textual form) from online social media websites.

Fakebuster: Fake News Detection System Using Logistic Regression Technique In Machine Learning

The uncontrollable spread of fake news through the net is irresistible in this globalization era. Fake news dissemination cannot be tolerated as the bad impacts of it to the society is really worrying. Furthermore, this will lead to more significant problems and potential threat such as confusion, misconceptions, slandering and luring users to share provocative lies made from fabricated news through their social media to occur. Within Malaysia context, there is lack in platform for fake news detection in Malay language articles and most of Malaysians received news through their social messaging applications. Fake news can be certainly solved by the aid of artificial intelligence which includes machine learning algorithms. The objective of this project is to propose a fake news detection model using Logistic Regression, to evaluate the performance of Logistic Regression as fake news detection model and to develop a web application that allows entry of a news content or news URL. In this study, Logistic Regression was applied in detecting fake news. Model development methodology is referenced and followed in this project. Based on existing studies, Logistic Regression showed a good performance in classification task. In addition, stancedetection approach is added to improve the accuracy of the model performance. Based on analysis made, this model within stance detection approach yields an excellent accuracy using TF-IDF feature in constructing this fake news model. This model is then integrated with web service that accepts input either news URL or news content in text which is then checked for its truth level through “FAKEBUSTER” application.

Frequent itemset-based feature selection and Rider Moth Search Algorithm for document clustering

Document clustering has recently been paid great attention in retrieval, navigation, and summarization of huge volumes of documents. With a better document clustering approach, computers can organize a document corpus automatically to a meaningful cluster for enabling efficient navigation, and browsing of the corpus. Document navigation and browsing is a valuable complement to the deficiencies of information retrieval technologies. This paper introduces Modsup-based frequent itemset and Rider Optimization-based Moth Search Algorithm (Rn-MSA) for clustering the documents. At first, the input documents are given to the pre-processing step, and then, the extraction is carried out based on TF-IDF and Wordnet features. Once the extraction is done, the feature selection is carried out based on frequent itemset for the establishment of feature knowledge. At last, the document clustering is done using the proposed Rn-MSA, which is designed by combining Rider Optimization Algorithm (ROA), and the Moth Search Algorithm (MSA). The performance of the document clustering based on proposed Modsup + Rn-MSA is evaluated in terms of precision, recall, F-Measure, and accuracy. The developed document clustering method achieves the maximal precision of 95.90%, maximal recall of 96.41%, maximal F-Measure of 96.41%, and the maximal accuracy of 95.12% that indicates its superiority.

To the question of data-driven identification of author’s age for Russian texts with age deceptions using machine learning

In this work we compare data-driven approaches to the task of author’s age identification for Russian texts with age deception. The data corpus has been specially gathered with crowdsourcing for this task. Two ways to determine age deception in author texts are considered and compared: the first is a traditional task of identification of age group of a text author, the second is identification of the occurrence of age imitation in the text with its type (imitating higher age or imitating lower age). The best results obtained by LinearSVC model with vector of TF-IDF features of character n-grams as input data demonstrate the F1-score of about 80% for the second task, and for the first one it is about 44%.

A clinical text classification paradigm using weak supervision and deep representation

BackgroundAutomatic clinical text classification is a natural language processing (NLP) technology that unlocks information embedded in clinical narratives. Machine learning approaches have been shown to be effective for clinical text classification tasks. However, a successful machine learning model usually requires extensive human efforts to create labeled training data and conduct feature engineering. In this study, we propose a clinical text classification paradigm using weak supervision and deep representation to reduce these human efforts.MethodsWe develop a rule-based NLP algorithm to automatically generate labels for the training data, and then use the pre-trained word embeddings as deep representation features for training machine learning models. Since machine learning is trained on labels generated by the automatic NLP algorithm, this training process is called weak supervision. We evaluat the paradigm effectiveness on two institutional case studies at Mayo Clinic: smoking status classification and proximal femur (hip) fracture classification, and one case study using a public dataset: the i2b2 2006 smoking status classification shared task. We test four widely used machine learning models, namely, Support Vector Machine (SVM), Random Forest (RF), Multilayer Perceptron Neural Networks (MLPNN), and Convolutional Neural Networks (CNN), using this paradigm. Precision, recall, and F1 score are used as metrics to evaluate performance.ResultsCNN achieves the best performance in both institutional tasks (F1 score: 0.92 for Mayo Clinic smoking status classification and 0.97 for fracture classification). We show that word embeddings significantly outperform tf-idf and topic modeling features in the paradigm, and that CNN captures additional patterns from the weak supervision compared to the rule-based NLP algorithms. We also observe two drawbacks of the proposed paradigm that CNN is more sensitive to the size of training data, and that the proposed paradigm might not be effective for complex multiclass classification tasks.ConclusionThe proposed clinical text classification paradigm could reduce human efforts of labeled training data creation and feature engineering for applying machine learning to clinical text classification by leveraging weak supervision and deep representation. The experimental experiments have validated the effectiveness of paradigm by two institutional and one shared clinical text classification tasks.

MHLM Majority Voting Based Hybrid Learning Model for Multi-Document Summarization

Text summarization from multiple documents is an active research area in the current scenario as the data in the World Wide Web (WWW) is found in abundance. The text summarization process is time-consuming and hectic for the users to retrieve the relevant contents from this mass collection of the data. Numerous techniques have been proposed to provide the relevant information to the users in the form of the summary. Accordingly, this article presents the majority voting based hybrid learning model (MHLM) for multi-document summarization. First, the multiple documents are subjected to pre-processing, and the features, such as title-based, sentence length, numerical data and TF-IDF features are extracted for all the individual sentences of the document. Then, the feature set is sent to the proposed MHLM classifier, which includes the Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Neural Network (NN) classifiers for evaluating the significance of the sentences present in the document. These classifiers provide the significance scores based on four features extracted from the sentences in the document. Then, the majority voting model decides the significant texts based on the significance scores and develops the summary for the user and thereby, reduces the redundancy, increasing the quality of the summary similar to the original document. The experiment performed with the DUC 2002 data set is used to analyze the effectiveness of the proposed MHLM that attains the precision and recall at a rate of 0.94, f-measure at a rate of 0.93, and ROUGE-1 at a rate of 0.6324.

A Particle Swarm Optimized Learning Model of Fault Classification in Web-Apps

The term web-app defines the current dynamic pragmatics of the website, where the user has control. Finding faults in such dynamic content is challenging, as to whether the fault is exposed or not depends on its execution path. Moreover, the complexity and uniqueness of each web application make fault assessment an extremely laborious and expensive task. Also, artificial fault injection models are run in controlled and simulated environments, which may not be representative of the real-world fault data. Classifying faults can intelligently enhance the quality of the web-apps by the assessment of probable faults. In this paper, an empirical study is conducted to classify faults in bug reports of three open-source web-apps ( qaManager , bitWeaver , and WebCalendar ) and reviews of two play store web-apps ( Dineout: Reserve a Table and Wynk Music ). Five supervised learning algorithms (naive Bayesian, decision tree, support vector machines, $K$ -nearest neighbor, and multi-layer perceptron) have been first evaluated based on the conventional term frequency–inverse document frequency (tf-idf) feature extraction method, and subsequently, a feature selection method to improve classifier performance is proposed using particle swarm optimization (a nature-inspired, meta-heuristic algorithm). This paper is a preliminary exploratory study to build an automated tool, which can optimally categorize faults. The empirical analysis validates that the particle swarm optimization for feature selection in fault classification task outperforms the tf-idf filter-based classifiers with an average accuracy gain of about 11% and nearly 26% average feature reduction. The highest accuracy of 93.35% is shown by the decision tree after feature selection.

Review on sentiment analysis in Tamil texts

Sentiment Analysis (SA) is an application of Natural Language Processing (NLP) to analyse the sentiments expressed in the text. It classifies into categories of qualities and opinions such as good, bad, positive, negative, neutral, etc. It employs machine learning techniques and lexicons for the classification. Nowadays, people share their opinions or feelings about movies, products, services, etc. through social media and online review sites. Analysing their opinions is beneficial to the public, business organisations, film producers and others to make decisions and improvements. SA is mostly employed in English language but rare for Indian languages including Tamil. This review paper aims to critically analyse the recent literature in the field of SA with Tamil text. Objectives, Methodologies and success rates are taken in consideration for the review. We shall conclude from the review that SVM and RNN classifiers taking TF-IDF and Word2vec features of Tamil text give better performance than grammar rules based classifications and other classifiers with presence of words, TF and BoW as features.

Fault Classification Method for Power Dispatching Log Based on Text Mining

The construction of smart grid has inevitably become the direction of the development of the power grid, and the use of data mining technologies to improve the level of data analysis is an important part of the construction of smart grid. In power grid scheduling, as the basis for the dispatcher to record the operation status of the power grid, the scheduling log usually uses short text to record the current state of the power grid, accident handling and abnormal alarm. However, it is lack of further analysis of the log. Aiming at the automatic classification of power grid dispatching log faults, we firstly use the custom power word and word segmentation tool to preprocess the text. After obtaining the TF-IDF feature of the text and creating the Word2Vec feature model, we compare three kinds of text classification algorithms, the nearest neighbor algorithm, Naive Bayes and support vector machine respectively. It is found that the support vector machine method is very effective in classification with a correct rate of 93% in the log short text.

The Performance Comparison of the Classifiers According to Binary Bow, Count Bow and Tf-Idf Feature Vectors for Malware Detection

In this paper, we compared the performance of the classifiers according to feature vectors with Binary BOW, Count BOW and TF-IDF for malware detection. We used the feature of Opcode that extracted from PE file. For performance comparison, we measured the AUC score for the classifiers those are DT, KNN, MLP, MNB and SVM. As a result, we recommend neural network (MLP) and instance-based model (KNN) because they show the high AUC score and accuracy regardless of the unbalanced dataset and the feature vector. If you use classical classifiers, we recommend DT because it guarantees high AUC score and accuracy regardless of the same condition as the above. If you use SVM, you have to do Robust scaling to resolved outlier and unbalanced dataset. If you use MNB, you need to use N-gram technique to improve AUC score.

Anatomy of Online Hate: Developing a Taxonomy and Machine Learning Models for Identifying and Classifying Hate in Online News Media

Online social media platforms generally attempt to mitigate hateful expressions, as these comments can be detrimental to the health of the community. However, automatically identifying hateful comments can be challenging. We manually label 5,143 hateful expressions posted to YouTube and Facebook videos among a dataset of 137,098 comments from an online news media. We then create a granular taxonomy of different types and targets of online hate and train machine learning models to automatically detect and classify the hateful comments in the full dataset. Our contribution is twofold: 1) creating a granular taxonomy for hateful online comments that includes both types and targets of hateful comments, and 2) experimenting with machine learning, including Logistic Regression, Decision Tree, Random Forest, Adaboost, and Linear SVM, to generate a multiclass, multilabel classification model that automatically detects and categorizes hateful comments in the context of online news media. We find that the best performing model is Linear SVM, with an average F1 score of 0.79 using TF-IDF features. We validate the model by testing its predictive ability, and, relatedly, provide insights on distinct types of hate speech taking place on social media.

A novel feature extraction methodology for sentiment analysis of product reviews

Feature extraction is one of the key steps for text sentiment analysis (SA), and the corresponding algorithms have important effect on the results. In the paper, a novel methodology is proposed to extract the feature for SA of product reviews. First, based on the diversified expression forms of product reviews, the generalized TF–IDF feature vectors are obtained by introducing the semantic similarity of synonyms. Then, in view of the different lengths of product reviews, the local patterns of the feature vectors are identified with OPSM biclustering algorithm. Finally, we improve PrefixSpan algorithm to detect the frequent and pseudo-consecutive phrases with high discriminative ability (namely FPCD phrases), which contain word-order information. Furthermore, some important factors, such as the separation and discriminative ability of words, are also employed to improve the discriminative ability of sentiment polarity. Based on the previous steps, the text feature vectors are extracted. A series of the experiment and comparison results indicate that the performance for SA on product review is greatly improved.

Verbal aggression detection on Twitter comments: convolutional neural network for short-text sentiment analysis

Cyberbullying and hate speeches are common issues in online etiquette. To tackle this highly concerned problem, we propose a text classification model based on convolutional neural networks for the de facto verbal aggression dataset built in our previous work and observe significant improvement, thanks to the proposed 2D TF-IDF features instead of pre-trained methods. Experiments are conducted to demonstrate that the proposed system outperforms our previous methods and other existing methods. A case study of word vectors is carried out to address the difficulty in using pre-trained word vectors for our short-text classification task, demonstrating the necessities of introducing 2D TF-IDF features. Furthermore, we also conduct visual analysis on the convolutional and pooling layers of the convolutional neural networks trained.

A Supervised Bug Report Classification with Incorporate and Textual field Knowledge

Performance of the bug prediction model is directly depends on the misclassification of bug reports. Misclassification issue surely scarifies the accuracy of the system. To resolve this issue the manual examination of bug reports are required, but it is very time consuming and tedious job for a developer and tester. In this paper the hybrid approach of merging text mining, natural language processing and machine learning techniques is used to identify bug report as bug or non-bug. The four incorporates fields with textual fields are added to bug reports to improve the performance of classifier. TF-IDF and Bigram feature extraction methods are used with feature selection and K-nearest neighbor (K-NN) classifier. The performance of the proposed system is evaluated by using Precision, Recall and F-measure by using five datasets. It is observed that the performance of K-NN classifier is changed according to the dataset and addition of bigram method improve the performance of classifier.

Kannada Named Entity Recognition and Classification using Support Vector Machine

Named Entity Recognition and Classification (NERC) is a process of identification of proper nouns in the text and classification of those nouns into certain predefined categories like person name, location, organization, date, time etc. Kannada NERC is an essential and challenging work which aims at developing a novel model based on Support Vector Machine. In this paper, tf-idf and POS features are used, which are extracted from a training corpus created manually. Furthermore, the model is trained and tested with different kernels: polynomial, rbf, sigmoid and linear kernels. The details of implementation and performance evaluation are discussed. The experiments are conducted on a training corpus of size 1, 51,440 tokens and test corpus of 7,000, 11,000, 15,000, 20,000, 30,000, 40,000 and 50,000 tokens. It is observed that the model works with an average precision, recall and F1-measure of 87%, 88% and 87.5% respectively for a linear kernel SVM on the test corpus of 7,000 tokens.

Sentiment Classification of Financial News Using Statistical Features

Sentiment classification of financial news deals with the identification of positive and negative news so that they can be applied in decision support systems for stock trend predictions. This paper explores several types of feature spaces as different data spaces for sentiment classification of the news article. Experiments are conducted using [Formula: see text]-gram models unigram, bigram and the combination of unigram and bigram as feature extraction with traditional feature weighting methods (binary, term frequency (TF), and term frequency-document frequency (TF-IDF)), while document frequency (DF) was used in order to generate feature spaces with different dimensions to evaluate [Formula: see text]-gram models and traditional feature weighting methods. We performed some experiments to measure the classification accuracy of support vector machine (SVM) with two kernel methods of Linear and Gaussian radial basis function (RBF). We concluded that feature selection and feature weighting methods can have a substantial role in sentiment classification. Furthermore, the results showed that the proposed work which combined unigram and bigram along with TF-IDF feature weighting method and optimized RBF kernel SVM produced high classification accuracy in financial news classification.

Prescription Function Prediction Using Topic Model and Multilabel Classifiers.

Determining a prescription's function is one of the challenging problems in Traditional Chinese Medicine (TCM). In past decades, TCM has been widely researched through various methods in computer science, but none concentrates on the prediction method for a new prescription's function. In this study, two methods are presented concerning this issue. The first method is based on a novel supervised topic model named Label-Prescription-Herb (LPH), which incorporates herb-herb compatibility rules into learning process. The second method is based on multilabel classifiers built by TFIDF features and herbal attribute features. Experiments undertaken reveal that both methods perform well, but the multilabel classifiers slightly outperform LPH-based method. The prediction results can provide valuable information for new prescription discovery before clinical test.

Predictive mining for stock market based on live news TF-IDF features

The various machine learning algorithms are used for prediction of stock market movement. The data collected for stock market is in the form of breaking news from various finance websites. The TF-IDF features extracted from online news data are used for creation of HMM model along with log likelihood values. The next day's stock price is predicted as either higher or lower than current day's stock price. Results obtained from proposed model is compared with results from other machine learning predictive techniques such as random forest, KNN, multiple regression, bagging and boosting. The proposed model produces approximately 70% of accurate prediction. The captured features are graphically represented with word cloud. The results can be further improved with the use of deep learning ensemble methods.

Predictive mining for stock market based on live news TF-IDF features

Predictive mining for stock market based on live news TF-IDF features

Large-Scale Text Similarity Computing with Spark

Text understanding is a hot research in Natural Language Processing and Information Retrieval. In recent years, it has received wide attention and research. In the era of big data, Understanding text in large-scale datasets is a challenge. Although the earliest systems designed for these workloads, such as MapReduce, gave users a powerful, but low-level, procedural programming interface. So, MapReduce doesn’t compose well for lager text applications. Recently, Spark, an in-memory cluster-computing platform, has been proposed. It has emerged as a popular framework for large-scale data processing and analytics. It provides a general-purpose efficient cluster computing engine and simpler for the end users. In this work, we consider using Vector Space Model (VSM) and TF-IDF weighting schema and feature hashing feature extraction techniques in order to solve the problem of large-scale text data similarity computing by Spark. As a result, Experimental results that using Spark in order to solve document similarity computation problems as soon as quickly by 20Newsgroups. In additions, It is more benefit from document classification and clustering of machine learning tasks.