Oversampling Method Research Articles

Detecting Hypoglycemia Incidents Reported in Patients' Secure Messages: Using Cost-Sensitive Learning and Oversampling to Reduce Data Imbalance.

BackgroundImproper dosing of medications such as insulin can cause hypoglycemic episodes, which may lead to severe morbidity or even death. Although secure messaging was designed for exchanging nonurgent messages, patients sometimes report hypoglycemia events through secure messaging. Detecting these patient-reported adverse events may help alert clinical teams and enable early corrective actions to improve patient safety.ObjectiveWe aimed to develop a natural language processing system, called HypoDetect (Hypoglycemia Detector), to automatically identify hypoglycemia incidents reported in patients’ secure messages.MethodsAn expert in public health annotated 3000 secure message threads between patients with diabetes and US Department of Veterans Affairs clinical teams as containing patient-reported hypoglycemia incidents or not. A physician independently annotated 100 threads randomly selected from this dataset to determine interannotator agreement. We used this dataset to develop and evaluate HypoDetect. HypoDetect incorporates 3 machine learning algorithms widely used for text classification: linear support vector machines, random forest, and logistic regression. We explored different learning features, including new knowledge-driven features. Because only 114 (3.80%) messages were annotated as positive, we investigated cost-sensitive learning and oversampling methods to mitigate the challenge of imbalanced data.ResultsThe interannotator agreement was Cohen kappa=.976. Using cross-validation, logistic regression with cost-sensitive learning achieved the best performance (area under the receiver operating characteristic curve=0.954, sensitivity=0.693, specificity 0.974, F1 score=0.590). Cost-sensitive learning and the ensembled synthetic minority oversampling technique improved the sensitivity of the baseline systems substantially (by 0.123 to 0.728 absolute gains). Our results show that a variety of features contributed to the best performance of HypoDetect.ConclusionsDespite the challenge of data imbalance, HypoDetect achieved promising results for the task of detecting hypoglycemia incidents from secure messages. The system has a great potential to facilitate early detection and treatment of hypoglycemia.

Generative Oversampling Method for Imbalanced Data on Bearing Fault Detection and Diagnosis

In this study, we developed a novel data-driven fault detection and diagnosis (FDD) method for bearing faults in induction motors where the fault condition data are imbalanced. First, we propose a bearing fault detector based on convolutional neural networks (CNN), in which the vibration signals from a test bench are used as inputs after an image transformation procedure. Experimental results demonstrate that the proposed classifier for FDD performs well (accuracy of 88% to 99%) even when the volume of normal and fault condition data is imbalanced (imbalance ratio varies from 20:1 to 200:1). Additionally, our generative model reduces the level of data imbalance by oversampling. The results improve the accuracy of FDD (by up to 99%) when a severe imbalance ratio (200:1) is assumed.

Synthetic Minority Over-Sampling for Improving Imbalanced Data in Educational Web Usage Mining

Educational data mining is the method for extracting and discovering new knowledge from education data. As education data is often complex and imbalanced, it requires a data preprocessing step or learning algorithms in order to obtain accurate analysis and interpretation. Many studies emphasize on classification and clustering methods in order to get insight and comprehensive knowledge from education data. However, a small number of previous works exclusively focused on the preprocessing of education data, particularly on the topic of the imbalanced dataset. Therefore, this research objective is to enhance the accuracy of data classification in educational web usage data. Our study involves the application of synthetic minority over-sampling techniques (SMOTE) to preprocess the raw dataset from web usage data. The minority class is a group of the students who failed the examination and the majority class is the students who passed the examination. In our experiments, four synthetic minority over-sampling methods are applied, SMOTE, and its variants: Borderline-SMOTE1, Borderline-SMOTE2, and SVM-SMOTE, in order to balance the number of samples in the minority class. The experiments are evaluated by comparing the results from well-known classification methods that are Naive Bayesian, decision tree, and k-nearest neighbors. The study experiments with real-world datasets from education data. The results present that synthetic minority over-sampling methods are capable of improving the detection of the minority class and achieve improving classification performance on precision, recall, and F1-value.
 Ed

Boosting the performance of over-sampling algorithms through under-sampling the minority class

Over-sampling algorithms are the most adopted approach to balance class distribution in imbalanced data problems, through random replication or synthesis of new examples in the minority class. Current over-sampling algorithms, however, usually use all available examples in the minority class to synthesise new instances, which may include noisy or outlier data. This work proposes k-INOS, a new algorithm to prevent over-sampling algorithms from being contaminated by noisy examples in the minority class. k-INOS is based on the concept of neighbourhood of influence and works as a wrapper around any over-sampling algorithm. A comprehensive experimentation was conducted to test k-INOS in 50 benchmark data sets, 8 over-sampling methods and 5 classifiers, with performance measured according to 7 metrics and Wilcoxon signed-ranks test. Results showed, particularly for weak classifiers (but not only), k-INOS significantly improved the performance of over-sampling algorithms in most performance metrics. Further investigations also allowed to identify conditions where k-INOS is likely to increase performance, according to features and rates measured from the data sets. The extensive experimentation framework evidenced k-INOS as an efficient algorithm to be applied prior to over-sampling methods.

Radial-Based oversampling for noisy imbalanced data classification

Imbalanced data classification remains a focus of intense research, mostly due to the prevalence of data imbalance in various real-life application domains. A disproportion among objects from different classes may significantly affect the performance of standard classification models. The first problem is the high imbalance ratios that pose a serious learning difficulty and require usage of dedicated methods, capable of alleviating this issue. The second important problem which may appear is noise, which may be accompanying the training data and causing strong deterioration of the classifier performance or increase the time required for its training. Therefore, the desirable classification model should be robust to both skewed data distributions and noise. One of the most popular approaches for handling imbalanced data is oversampling of the minority objects in their neighborhood. In this work we will criticize this approach and propose a novel strategy for dealing with imbalanced data, with particular focus on the noise presence. We propose Radial-Based Oversampling (RBO) method, which can find regions in which the synthetic objects from minority class should be generated on the basis of the imbalance distribution estimation with radial basis functions. Results of experiments, carried out on a representative set of benchmark datasets, confirm that the proposed guided synthetic oversampling algorithm offers an interesting alternative to popular state-of-the-art solutions for imbalanced data preprocessing.

Tackling Class Imbalance Problem in Software Defect Prediction Through Cluster-Based Over-Sampling With Filtering

In practice, Software Defect Prediction (SDP) models often suffer from highly imbalanced data, which makes classifiers difficult to identify defective instances. Recently, many techniques were proposed to tackle this problem, over-sampling technique is one of the most well-known methods to address class imbalance problem. This technique balances the number of defective and non-defective instances by generating new defective instances. However, these approaches would generate non-diverse synthetic instances, and many unnecessary noise instances at the same time. Motived by this, we propose a Cluster-based Over-sampling with noise filtering (KMFOS) approach to tackle class imbalance problem in SDP. KMFOS firstly divides defective instances into $K$ clusters, and new defective instances are generated by interpolation between instances of each two clusters. After this, these new defective instances would diversely spread in the space of defective dataset. Then, we extend this cluster-based over-sampling through the Closest List Noise Identification (CLNI) to clean the noise instances. We do extensive experiments on 24 projects to compare KMFOS with some over-sampling approaches such as SMOTE, Borderline-SMOTE, ADASYN, random over-sampling (ROS), K-means SMOTE, SMOTE + IPF, SMOTE + ENN and SMOTE + Tomek Links using five prediction classifiers. At the same time, we also compare KMFOS with other state-of-the-art class-imbalance methods including balancebaggingclassifier, RUSboostclassifier, InstanceHardnessThreshold and cost-sensitive methods. Experimental results indicate our KMFOS can obtain better Recall and bal values than other over-sampling methods and other compared class-imbalance methods. Hence, KMFOS is an efficient approach to generate balanced data for SDP and improves the performance of predicting models.

Visualization and Analysis in Bank Direct Marketing Prediction

Gaining the most benefits out of a certain data set is a difficult task because it requires an in-depth investigation into its different features and their corresponding values. This task is usually achieved by presenting data in a visual format to reveal hidden patterns. In this study, several visualization techniques are applied to a bank’s direct marketing data set. The data set obtained from the UCI machine learning repository website is imbalanced. Thus, some oversampling methods are used to enhance the accuracy of the prediction of a client’s subscription to a term deposit. Visualization efficiency is tested with the oversampling techniques’ influence on multiple classifier performance. Results show that the agglomerative hierarchical clustering technique outperforms other oversampling techniques and the Naive Bayes classifier gave the best prediction results.

DeepBalance: Deep-Learning and Fuzzy Oversampling for Vulnerability Detection

Software vulnerability has long been an important but critical research issue in cybersecurity. Recently, the machine learning (ML)-based approach has attracted increasing interest in the research of software vulnerability detection. However, the detection performance of existing ML-based methods require further improvement. There are two challenges: one is code representation for ML and the other is class imbalance between vulnerable code and nonvulnerable code. To overcome these challenges, this article develops a DeepBalance system, which combines the new ideas of deep code representation learning and fuzzy-based class rebalancing. We design a deep neural network with bidirectional long short-term memory to learn invariant and discriminative code representations from labeled vulnerable and nonvulnerable code. Then, a new fuzzy oversampling method is employed to rebalance the training data by generating synthetic samples for the class of vulnerable code. To evaluate the performance of the new system, we carry out a series of experiments in a real-world ground-truth dataset that consists of the code from the projects of LibTIFF, LibPNG, and FFmpeg. The results show that the proposed new system can significantly improve the vulnerability detection performance. For example, the improvement is 15% in terms of F-measure.

Ensemble Feature Learning to Identify Risk Factors for Predicting Secondary Cancer.

Background: In recent years, the development and diagnosis of secondary cancer have become the primary concern of cancer survivors. A number of studies have been developing strategies to extract knowledge from the clinical data, aiming to identify important risk factors that can be used to prevent the recurrence of diseases. However, these studies do not focus on secondary cancer. Secondary cancer is lack of the strategies for clinical treatment as well as risk factor identification to prevent the occurrence.Methods: We propose an effective ensemble feature learning method to identify the risk factors for predicting secondary cancer by considering class imbalance and patient heterogeneity. We first divide the patients into some heterogeneous groups based on spectral clustering. In each group, we apply the oversampling method to balance the number of samples in each class and use them as training data for ensemble feature learning. The purpose of ensemble feature learning is to identify the risk factors and construct a diagnosis model for each group. The importance of risk factors is measured based on the properties of patients in each group separately. We predict secondary cancer by assigning the patient to a corresponding group and based on the diagnosis model in this corresponding group.Results: Analysis of the results shows that the decision tree obtains the best results for predicting secondary cancer in the three classifiers. The best results of the decision tree are 0.72 in terms of AUC when dividing the patients into 15 groups, 0.38 in terms of F1 score when dividing the patients into 20 groups. In terms of AUC, decision tree achieves 67.4% improvement compared to using all 20 predictor variables and 28.6% improvement compared to no group division. In terms of F1 score, decision tree achieves 216.7% improvement compared to using all 20 predictor variables and 80.9% improvement compared to no group division. Different groups provide different ranking results for the predictor variables.Conclusion: The accuracies of predicting secondary cancer using k-nearest neighbor, decision tree, support vector machine indeed increased after using the selected important risk factors as predictors. Group division on patients to predict secondary cancer on the separated models can further improve the prediction accuracies. The information discovered in the experiments can provide important references to the personality and clinical symptom representations on all phases of guide interventions, with the complexities of multiple symptoms associated with secondary cancer in all phases of the recurrent trajectory.

Imbalanced Fault Diagnosis of Rolling Bearing Based on Generative Adversarial Network: A Comparative Study

Due to the real working conditions and data acquisition equipment, the collected working data of bearings are actually limited. Meanwhile, as the rolling bearing works in the normal state at most times, it is easy to raise the imbalance problem of fault types which restricts the diagnosis accuracy and stability. To solve these problems, we present an imbalanced fault diagnosis method based on the generative adversarial network (GAN) and provide a comparative study in detail. The key idea is utilizing GAN, a kind of deep learning technique, to generate synthetic samples for minority fault class and then improve the generalization ability of the fault diagnosis model. First, this method applies fast Fourier transform to pre-process the original vibration signal and then obtains the frequency spectrum of fault samples. Second, it uses the spectrum data as the input of GAN to generate the synthetic minority samples following the data distribution of the real samples. Finally, it puts the synthetic samples into the training set and builds a stacked denoising auto encoder model for fault diagnosis. To testify the effectiveness of the proposed method, a series of comparative experiments is carried out on the CWRU bearing dataset. The results show that the proposed method can provide a better solution for imbalanced fault diagnosis on the basis of generating similar fault samples. As a comparative study, the proposed method is compared to several diagnostic methods with traditional time-frequency domain characteristics. Moreover, we also demonstrate that the proposed method outperforms three widely used sample synthesis techniques, such as random oversampling, synthetic minority oversampling technique, and the principal curve-based oversampling method in terms of diagnosis accuracy and numerical stability.

A Parameter-Free Cleaning Method for SMOTE in Imbalanced Classification

Oversampling is an efficient technique in dealing with class-imbalance problem. It addresses the problem by reduplicating or generating the minority class samples to balance the distribution between the samples of the majority and the minority class. Synthetic minority oversampling technique (SMOTE) is one of the typical representatives. During the past decade, researchers have proposed many variants of SMOTE. However, the existing oversampling methods may generate wrong minority class samples in some scenarios. Furthermore, how to effectively mine the inherent complex characteristics of imbalanced data remains a challenge. To this end, this paper proposes a parameter-free data cleaning method to improve SMOTE based on constructive covering algorithm. The dataset generated by SMOTE is first partitioned into a group of covers, then the hard-to-learn samples can be detected based on the characteristics of sample space distribution. Finally, a pair-wise deletion strategy is proposed to remove the hard-to-learn samples. The experimental results on 25 imbalanced datasets show that our proposed method is superior to the comparison methods in terms of various metrics, such as ${F}$ -measure, ${G}$ -mean, and Recall. Our method not only can reduce the complexity of the dataset but also can improve the performance of the classification model.

An alternative SMOTE oversampling strategy for high-dimensional datasets

In this work, the Synthetic Minority Over-sampling Technique (SMOTE) approach is adapted for high-dimensional binary settings. A novel distance metric is proposed for the computation of the neighborhood for each minority sample, which takes into account only a subset of the available attributes that are relevant for the task. Three variants for the distance metric are explored: Euclidean, Manhattan, and Chebyshev distances, and four different ranking strategies: Fisher Score, Mutual Information, Eigenvector Centrality, and Correlation Score. Our proposal was compared with various oversampling techniques on low- and high-dimensional datasets with the presence of class-imbalance, including a case study on Natural Language Processing (NLP). The proposed oversampling strategy showed superior results on average when compared with SMOTE and other variants, demonstrating the importance of selecting the right attributes when defining the neighborhood in SMOTE-based oversampling methods.

SMOTE-BD: An Exact and Scalable Oversampling Method for Imbalanced Classification in Big Data

The volume of data in today's applications has meant a change in the way Machine Learning issues are addressed. Indeed, the Big Data scenario involves scalability constraints that can only be achieved through intelligent model design and the use of distributed technologies. In this context, solutions based on the Spark platform have established themselves as a de facto standard.
 In this contribution, we focus on a very important framework within Big Data Analytics, namely classification with imbalanced datasets. The main characteristic of this problem is that one of the classes is underrepresented, and therefore it is usually more complex to find a model that identifies it correctly. For this reason, it is common to apply preprocessing techniques such as oversampling to balance the distribution of examples in classes.
 In this work we present SMOTE-BD, a fully scalable preprocessing approach for imbalanced classification in Big Data. It is based on one of the most widespread preprocessing solutions for imbalanced classification, namely the SMOTE algorithm, which creates new synthetic instances according to the neighborhood of each example of the minority class. Our novel development is made to be independent of the number of partitions or processes created to achieve a higher degree of efficiency. Experiments conducted on different standard and Big Data datasets show the quality of the proposed design and implementation.

SMS Phishing Detection Using Oversampling and Feature Optimization Method

In this paper, a new SMS phishing detection method using oversampling and feature optimization technology is proposed to improve SMS phishing detection accuracy. Three types features are presented including token features, topic features and Linguistic Inquiry and Word Count (LIWC) features. One of the existing oversampling methods called Adaptive Synthetic Sampling Approach is applied in this paper since it has good performance. Then, Binary Particle Swarm Optimization(BPSO) algorithm is used to analyze the three types features and select the optimal combination of all the features. Finally, the detection results are achieved by Random Forest classification algorithm. Experimental results show oversampling method and feature optimization method improve the accuracy of SMS phishing detection. The best accuracy of the proposed method is 99.01% with an average of 86.6 features. The results demonstrate that the proposed method has a promising performance for SMS phishing detection.

A BOOTSTRAP APPROACH FOR IMPROVING LOGISTIC REGRESSION PERFORMANCE IN IMBALANCED DATA SETS

In an imbalanced dataset with binary response, the percentages of successes and failures are not approximately equal. In many real world situations, majority of the observations are “normal” (i.e., success) with a much smaller fraction of failures. The overall probability of correct classification for extremely imbalanced data sets can be very high but the probability of correctly predicting the minority class can be very low. Consider a fictitious example of a dataset with 1,000,000 observations out of which 999,000 are successes and 1,000 failures. A rule that classifies all observations as successes will have very high accuracy of prediction (99.9%) but the probability of correctly predicting a failure will be 0. In many situations, the cost associated with incorrect prediction of a failure is high, and it is therefore important to improve the prediction accuracy of failures as well. Literature suggests that over-sampling of the minority class with replacement does not necessarily predict the minority class with higher accuracy. In this article, we propose a simple over-sampling method which bootstraps a subset of the minority class, and illustrate the bootstrap over-sampling method with several examples. In each of these examples, an improvement in prediction accuracy is seen. Article DOI: https://dx.doi.org/10.20319/mijst.2018.43.1124 This work is licensed under the Creative Commons Attribution-Non-commercial 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc/4.0/ or send a letter to Creative Commons, PO Box 1866, Mountain View, CA 94042, USA.

A Comparative Study of Synthetic Over-sampling Method to Improve the Classification of Poor Households in Yogyakarta Province

The problems of class imbalance have attracted concerns from researchers in the last few years. Class imbalance problems occur when the data had unbalanced proportions between two or more groups of data which are usually called as minority and majority classes. These problems relate to creation of bias in parameter estimation as well as misclassification of the objects especially for the minority class. These will lead to incorrect prediction of the minority class, and eventually will risk the policy making. Several approaches have been proposed to correct misclassification such as data-based and algorithm-based approaches. As a data-based approach, over-sampling method is very popular nowadays. This approach is basically balancing the distribution of data through addition of synthetic data. This paper discusses the strategies of adding synthetic data in order to improve the accuracy of classification. Moreover, this paper also reviews several over sampling methods for class imbalanced problems. Specifically, the classification of poor households is illustrated by using the National Socio-Economic Survey (Susenas) data which has been stratified according to urban and rural areas. Finally, the K-Nearest Neighbor (KNN), Naïve Bayes, Support Vector Machine (SVM) and Generalized Linear Model (GLM) are employed to evaluate the classification performance by comparing the value of sensitivity and area under the ROC curve (AUC). The simulation result shows that there are bias on parameter estimation both on interception and on slope. The bias gets bigger as the data condition becomes more unbalanced and on small sample. Meanwhile, the classification accuracy will decrease with the decrement of probability (high imbalanced) value especially in the data with small sample. Decreased accuracy of classification mainly occurs in the minority class (sensitivity) and AUC. Based on the simulation result, it is clear that the synthetic over sampling approach can improve the accuracy of classification in minority class through increasing sensitivity value and AUC value. This occur at the small probability (unbalanced data). In line with the simulation results, the over sampling approach also shows the evident of improving the prediction of poor households in Yogyakarta Province. But on the other hand, it can also lead to decreased accuracy and specificity. However, further research is required to obtain a more accurate prediction result for all performance measures.

A survey on addressing high-class imbalance in big data

In a majority–minority classification problem, class imbalance in the dataset(s) can dramatically skew the performance of classifiers, introducing a prediction bias for the majority class. Assuming the positive (minority) class is the group of interest and the given application domain dictates that a false negative is much costlier than a false positive, a negative (majority) class prediction bias could have adverse consequences. With big data, the mitigation of class imbalance poses an even greater challenge because of the varied and complex structure of the relatively much larger datasets. This paper provides a large survey of published studies within the last 8 years, focusing on high-class imbalance (i.e., a majority-to-minority class ratio between 100:1 and 10,000:1) in big data in order to assess the state-of-the-art in addressing adverse effects due to class imbalance. In this paper, two techniques are covered which include Data-Level (e.g., data sampling) and Algorithm-Level (e.g., cost-sensitive and hybrid/ensemble) Methods. Data sampling methods are popular in addressing class imbalance, with Random Over-Sampling methods generally showing better overall results. At the Algorithm-Level, there are some outstanding performers. Yet, in the published studies, there are inconsistent and conflicting results, coupled with a limited scope in evaluated techniques, indicating the need for more comprehensive, comparative studies.

Identifying Collapsed Buildings Using Post-Earthquake Satellite Imagery and Convolutional Neural Networks: A Case Study of the 2010 Haiti Earthquake

Earthquake is one of the most devastating natural disasters that threaten human life. It is vital to retrieve the building damage status for planning rescue and reconstruction after an earthquake. In cases when the number of completely collapsed buildings is far less than intact or less-affected buildings (e.g., the 2010 Haiti earthquake), it is difficult for the classifier to learn the minority class samples, due to the imbalance learning problem. In this study, the convolutional neural network (CNN) was utilized to identify collapsed buildings from post-event satellite imagery with the proposed workflow. Producer accuracy (PA), user accuracy (UA), overall accuracy (OA), and Kappa were used as evaluation metrics. To overcome the imbalance problem, random over-sampling, random under-sampling, and cost-sensitive methods were tested on selected test A and test B regions. The results demonstrated that the building collapsed information can be retrieved by using post-event imagery. SqueezeNet performed well in classifying collapsed and non-collapsed buildings, and achieved an average OA of 78.6% for the two test regions. After balancing steps, the average Kappa value was improved from 41.6% to 44.8% with the cost-sensitive approach. Moreover, the cost-sensitive method showed a better performance on discriminating collapsed buildings, with a PA value of 51.2% for test A and 61.1% for test B. Therefore, a suitable balancing method should be considered when facing imbalance dataset to retrieve the distribution of collapsed buildings.

A new MapReduce associative classifier based on a new storage format for large-scale imbalanced data

The process of knowledge discovery from big and high dimensional datasets has become a popular research topic. The classification problem is a key task in bioinformatics, business intelligence, decision science, astronomy, physics, etc. Building associative classifiers has been a notable research interest in recent years because of their superior accuracy. In associative classifiers, using under-sampling or over-sampling methods for imbalanced big datasets reduces accuracy or increases running time, respectively. Hence, there is a significant need to create efficient associative classifiers for imbalanced big data problems. These classifiers should be able to handle challenges such as memory usage, running time and efficiently exploring the search space. To this end, efficient calculation of measures is a primary objective for associative classifiers. In this paper, we propose a new efficient associative classifier for big imbalanced datasets. The proposed method is based on Rare-PEARs (a multi-objective evolutionary algorithm that efficiently discovers rare and reliable association rules) and is able to evaluate rules in a distributed manner by using a new storing data format. This format simplifies measures calculation and is fully compatible with the MapReduce programming model. We have applied the proposed method (RPII) on a well-known big dataset (ECBDL’14) and have compared our results with seven other learning methods. The experimental results show that RPII outperform other methods in sensitivity and final score measures (the values of sensitivity and final score measures were approximately 0.74 and 0.54 respectively). The results demonstrate that the proposed method is a good candidate for large-scale classification problems; furthermore, it achieves reasonable execution time when the target platform is a typical computer clusters.

Imbalanced text sentiment classification using universal and domain-specific knowledge

In this paper, a sentiment classification model is proposed to address two predominant issues in sentiment classification, namely domain-sensitive and data imbalance. Since words may embed distinct sentiment polarities in different contexts, sentiment classification is widely contended as a domain-sensitive task. Accordingly, this paper draws on label propagation to induce universal and domain-specific sentiment lexicons and builds a domain-adaptive sentiment classification model that incorporates universal and domain-specific knowledge into a unified learning framework. On the flip side, sentiment-related corpuses are usually formed with skewed polarity distribution because individuals tend to share similar assessment criteria on a given object and hence their sentiment polarities toward the same object are likely to be similar. We endeavor to address such imbalanced data problem by advancing a novel over-sampling technique. Unlike existing over-sampling approaches that generate minority-class samples from numerical feature space, the proposed sampling method directly creates synthetic texts from word spaces. Several experiments are conducted to verify the effectiveness of the proposed lexicon generation method, learning framework, and over-sampling method. Results show that the induced sentiment lexicons are interpretable and the proposed model is found to be effective for imbalanced and domain-specific text sentiment classification.