Sentiment Analysis Datasets Research Articles

SKEAFN: Sentiment Knowledge Enhanced Attention Fusion Network for multimodal sentiment analysis

Multimodal sentiment analysis is an active research field that aims to recognize the user’s sentiment information from multimodal data. The primary challenge in this field is to develop a high-quality fusion framework that effectively addresses the heterogeneity among different modalities. However, prior research has primarily concentrated on intermodal interactions while neglecting the semantic sentiment information conveyed by words in the text modality. In this paper, we propose the Sentiment Knowledge Enhanced Attention Fusion Network (SKEAFN), a novel end-to-end fusion network that enhances multimodal fusion by incorporating additional sentiment knowledge representations from an external knowledge base. Firstly, we construct an external knowledge enhancement module to acquire additional representations for the text modality. Then, we design a text-guided interaction module that facilitates the interaction between text and the visual/acoustic modality. Finally, we propose a feature-wised attention fusion module that achieves multimodal fusion by dynamically adjusting the weights of the additional and each modality’s representations. We evaluate our method on three challenging multimodal sentiment analysis datasets: CMU-MOSI, CMU-MOSEI, and Twitter2019. The experiment results demonstrate that our model significantly outperforms the state-of-the-art models. The source code is publicly available at https://github.com/doubibobo/SKEAFN.

A modified Bayesian boosting algorithm with weight-guided optimal feature selection for sentiment analysis

Sentiment analysis is crucial in understanding and analyzing public opinions, feedback, and social media data. In this study, we propose a modified Bayesian Boosting algorithm with weight-guided optimal feature selection for sentiment analysis. The goal is to improve the accuracy and efficiency of sentiment classification by selecting the most informative features and utilizing an optimized boosting algorithm. The proposed algorithm leverages the strength of the Bayesian framework and boosting techniques to handle the challenges of sentiment analysis effectively. It incorporates a feature selection mechanism that assigns weights to individual features based on their importance in sentiment classification. This weight-guided feature selection approach helps identify the most relevant and discriminative features, enhancing classification performance. The algorithm employs a modified Bayesian boosting algorithm to optimize the boosting process. It dynamically adjusts the weights assigned to weak classifiers during the boosting iterations, emphasizing the misclassified instances and providing more focused learning for accurate sentiment prediction. The proposed algorithm is evaluated on a benchmark sentiment analysis dataset, comparing it with other state-of-the-art algorithms. The results demonstrate that the modified Bayesian Boosting algorithm with weight-guided optimal feature selection achieves superior accuracy, precision, recall, and F1 score. The algorithm shows its effectiveness in capturing the sentiment patterns and improving the overall sentiment classification performance. The results demonstrated that the proposed model had gained an accuracy level of 98.49%, even the Random Forest algorithm gained an accuracy level of 78.6%, and the C4.5 model achieved an accuracy level of 68.58% individually. The proposed modified Bayesian Boosting algorithm with weight-guided optimal feature selection presents a promising approach for sentiment analysis. It effectively addresses the challenges of feature selection and boosting, leading to improved sentiment classification accuracy. The algorithm’s ability to adjust weights and select informative features adaptively enhances its sentiment analysis tasks performance. This research opens up new avenues for advancing sentiment analysis techniques and contributes to developing more accurate and efficient sentiment analysis models.

PerceptSent - Exploring Subjectivity in a Novel Dataset for Visual Sentiment Analysis

Visual sentiment analysis is a challenging problem. Many datasets and approaches have been designed to foster breakthroughs in this trending research topic. However, most works scrutinize only subsymbolic models through visual attributes of the evaluated images, paying less attention to the subjectivity of viewers' perceptions as a basis for neuro-symbolic systems. Aiming to fill this gap, we present PerceptSent, a novel dataset for visual sentiment analysis that spans 5,000 images shared by users on social networks. Besides the sentiment opinion (positive, slightly positive, neutral, slightly negative, negative) expressed by every evaluator about each image analyzed, the dataset contains evaluator's metadata (age, gender, socioeconomic status, education, and psychological hints) as well as perceptions observed by the evaluator about the image — such as the presence of nature, violence, lack of maintenance, etc. Deep architectures and different problem formulations are explored using our dataset to combine visual and extra attributes (external knowledge) for automatic sentiment analysis. We show evidence that evaluator's perceptionss, when correctly employed, are crucial in visual sentiment analysis, improving the F-score performance from 61% to an impressive rate above 97%. Although, at this point, we do not have automatic approaches to capture these perceptions, our results open up new investigation avenues.

Resource Construction and Ensemble Learning based Sentiment Analysis for the Low-resource Language Uyghur

<p>To address the problem of scarce low-resource sentiment analysis corpus nowadays, this paper proposes a sentence-level sentiment analysis resource conversion method HTL based on the syntactic-semantic knowledge of the low-resource language Uyghur to convert high-resource corpus to low-resource corpus. In the conversion process, a k-fold cross-filtering method is proposed to reduce the distortion of data samples, which is used to select high-quality samples for conversion; finally, the Uyghur sentiment analysis dataset USD is constructed; the Baseline of this dataset is verified under the LSTM model, and the accuracy and F1 values reach 81.07% and 81.13%, respectively, which can provide a reference for the construction of low-resource language corpus nowadays. The accuracy and F1 values reached 81.07% and 81.13%, respectively, which can provide a reference for the construction of today’s low-resource corpus. Meanwhile, this paper also proposes a sentiment analysis model based on logistic regression ensemble learning, SA-LREL, which combines the advantages of several lightweight network models such as TextCNN, RNN, and RCNN as the base model, and the meta-model is constructed using logistic regression functions for ensemble, and the accuracy and F1 values reach 82.17% and 81.86% respectively in the test set, and the experimental results show that the method can effectively improve the performance of Uyghur sentiment analysis task.</p> <p> </p>

Vietnamese Sentiment Analysis: An Overview and Comparative Study of Fine-tuning Pretrained Language Models

Sentiment Analysis (SA) is one of the most active research areas in the Natural Language Processing (NLP) field due to its potential for business and society. With the development of language representation models, numerous methods have shown promising efficiency in fine-tuning pre-trained language models in NLP downstream tasks. For Vietnamese, many available pre-trained language models were also released, including the monolingual and multilingual language models. Unfortunately, all of these models were trained on different architectures, pre-trained data, and pre-processing steps; consequently, fine-tuning these models can be expected to yield different effectiveness. In addition, there is no study focusing on evaluating the performance of these models on the same datasets for the SA task up to now. This article presents a fine-tuning approach to investigate the performance of different pre-trained language models for the Vietnamese SA task. The experimental results show the superior performance of the monolingual PhoBERT model and ViT5 model in comparison with previous studies and provide new state-of-the-art performances on five benchmark Vietnamese SA datasets. To the best of our knowledge, our study is the first attempt to investigate the performance of fine-tuning Transformer-based models on five datasets with different domains and sizes for the Vietnamese SA task.

A comparison of multiple word embeddings and performance analysis

Based on IMDB data sets for sentiment analysis, this research evaluates the performance of two types of neural networks, namely FNN and BERT. According to the experimental findings, both varieties of neural networks performed well (over 80%) on IMDB data sets. BERT performs the best of them all. The version of FNN improves when the number of layers is increased. The causes behind BERT's outstanding performance are then further examined in this research. In this article, it is hypothesized that a few factors, such as feature extraction capability and classification capability, account for BERT's exceptional performance. The results show that for the two coupling variables of BERT assumed in this paper, including feature extraction and classification ability, BERT has obvious advantages in feature extracting, and its classification ability has certain advantages over SVM and MLP. The deep learning model has achieved excellent performance in the sentiment classification task. The next step of this paper will focus on the robustness of the BERT model in sentiment analysis.

Implicit sentiment analysis based on affective knowledge and event information

In recent years, implicit sentiment analysis, which aims to detect sentiment of a sentence that don’t contain obvious sentiment words, has become an attractive research topic. This paper focuses on event-centric implicit sentiment analysis, which utilizes the sentiment-aware event to infer the sentiment polarity of the sentence. Existing event-based implicit sentiment analysis methods typically treat entities or noun phrases in the text as events, or model contextual information to indirectly infer events using sophisticated models, but these methods fail to fully capture event information. To address these issues, this paper defines events as <subject, predicate, object>. Based on this event representation, neural tensor network was used to model the interaction between event elements and extract high-level semantic features of events. In addition, a novel affective enhanced graph model was proposed to capture sentiment-related dependencies between context words. Furthermore, this paper considers the case where a sentence contains multiple events, and constructs an event-centric implicit sentiment analysis dataset, where each sentence contains at least one event triplet. Experimental results on the constructed dataset demonstrate the effectiveness of our proposed approach.

SRL-ACO: A text augmentation framework based on semantic role labeling and ant colony optimization

The process of creating high-quality labeled data is crucial for training machine-learning models, but it can be a time-consuming and labor-intensive process. Moreover, manual annotation by human annotators can lead to varying degrees of competency, training, and experience, which can result in inconsistent labeling and arbitrary standards. To address these challenges, researchers have been exploring automated methods for enhancing training and testing datasets. This paper proposes SRL-ACO, a novel text augmentation framework that leverages Semantic Role Labeling (SRL) and Ant Colony Optimization (ACO) techniques to generate additional training data for natural language processing (NLP) models. The framework uses SRL to identify the semantic roles of words in a sentence and ACO to generate new sentences that preserve these roles. SRL-ACO can enhance the accuracy of NLP models by generating additional data without requiring manual data annotation. The paper presents experimental results demonstrating the effectiveness of SRL-ACO on seven text classification datasets for sentiment analysis, toxic text detection and sarcasm identification. The results show that SRL-ACO improves the performance of a classifier on different NLP tasks. These results demonstrate that SRL-ACO has the potential to enhance the quality and quantity of training data for various NLP tasks.

Implicit Emotion Analysis Based on Improved Supervised Contrastive Learning

Implicit sentiment analysis (ISA) can be distinguished from traditional text sentiment analysis by the fact that it does not rely on emotional words as emotional clues, and the expression is usually vaguer. Identifying implicit emotions is more difficult, as it requires a deeper understanding of the context, even when emotional words are absent. Researchers have focused on context feature modeling and developing sophisticated feature extraction mechanisms instead of starting from the emotional perspective. Enhancing the difference in emotional features of text samples is an intuitive method to address this challenge. We proposed a supervised contrastive learning (SCL) method during training that enables the model to conduct contrastive learning based on emotion labels even while training on weak emotion features. SCL training can strengthen the average embedding distance between texts with different emotion labels and enhance implicit emotion discrimination. Moreover, research indicates that contextual information can improve implicit emotion classification ability. Therefore, we applied a straightforward context feature fusion method (bi-affine) over a more complicated context feature modeling approach. To evaluate the effectiveness of our proposed method, we conducted experiments on the SMP2019-ECISA (Chinese implicit sentiment analysis) dataset. The results show a 2.13% enhancement in the F1 value compared to the BERT baseline, proving the effectiveness of our methods.

Natural Language Processing and Sentiment Analysis on Bangla Social Media Comments on Russia–Ukraine War Using Transformers

The Bangla Language ranks seventh in the list of most spoken languages with 265 native and non-native speakers around the world and the second Indo-Aryan language after Hindi. However, the growth of research for tasks such as sentiment analysis (SA) in Bangla is relatively low compared to SA in the English language. It is because there are not enough high-quality publically available datasets for training language models for text classification tasks in Bangla. In this paper, we propose a Bangla annotated dataset for sentiment analysis on the ongoing Ukraine–Russia war. The dataset was developed by collecting Bangla comments from various videos of three prominent YouTube TV news channels of Bangladesh covering their report on the ongoing conflict. A total of 10,861 Bangla comments were collected and labeled with three polarity sentiments, namely Neutral, Pro-Ukraine (Positive), and Pro-Russia (Negative). A benchmark classifier was developed by experimenting with several transformer-based language models all pre-trained on unlabeled Bangla corpus. The models were fine-tuned using our procured dataset. Hyperparameter optimization was performed on all 5 transformer language models which include: BanglaBERT, XLM-RoBERTa-base, XLM-RoBERTa-large, Distil-mBERT and mBERT. Each model was evaluated and analyzed using several evaluation metrics which include: F1 score, accuracy, and AIC (Akaike Information Criterion). The best-performing model achieved the highest accuracy of 86% with 0.82 F1 score. Based on accuracy, F1 score and AIC, BanglaBERT outperforms baseline and all the other transformer-based classifiers.

Multimodal Emotion Analysis Model based on Interactive Attention Mechanism

In traditional multi-modal sentiment analysis, feature fusion is usually achieved by simple splicing, and multi-modal sentiment analysis is only trained as a single task, without considering the contribution of inter-modal information interaction to sentiment analysis and the correlation and constraint relationship between multi-modal and single-modal (text, video and audio) tasks. Therefore, a multi-task model based on interactive attention mechanism is proposed in this paper, which uses inter-modal attention mechanism and single-modal self-attention mechanism to train multi-modal sentiment analysis and single-modal sentiment analysis together, so as to make full use of inter-modal and inter-task information sharing, mutual complement, and reduce noise to improve the overall recognition performance. Experiments show that the proposed model performs well on MOSI and MOSEI common data sets for multimodal sentiment analysis.

Lightweight multilayer interactive attention network for aspect-based sentiment analysis

Aspect-based sentiment analysis (ABSA) aims to automatically identify the sentiment polarity of specific aspect words in a given sentence or document. Existing studies have recognised the value of interactive learning in ABSA and have developed various methods to precisely model aspect words and their contexts through interactive learning. However, these methods mostly take a shallow interactive way to model aspect words and their contexts, which may lead to the lack of complex sentiment information. To solve this issue, we propose a Lightweight Multilayer Interactive Attention Network (LMIAN) for ABSA. Specifically, we first employ a pre-trained language model to initialise word embedding vectors. Second, an interactive computational layer is designed to build correlations between aspect words and their contexts. Such correlation degree is calculated by multiple computational layers with neural attention models. Third, we use a parameter-sharing strategy among the computational layers. This allows the model to learn complex sentiment features with lower memory costs. Finally, LMIAN conducts instance validation on six publicly available sentiment analysis datasets. Extensive experiments show that LMIAN performs better than other advanced methods with relatively low memory consumption.

Alternative Text Pre-Processing using Chat GPT Open AI

Text Pre-Processing is the first step in Sentiment Analysis. Categorizing a sentiment in a dataset is part of the Text-Preprocessing stage to get the optimal model accuracy value. Generative Pretrained Transformer, often known as Chat GPT, is a Machine Learning model that can automatically generate realistic and meaningful text. This study aims to examine the capability of GPT Chat as an alternative in the Text-Pre-Processing stage by utilizing GPT Chat 3 from the openai.com website in the Text-Pre-Processing stage of the collected tweet data. The data used in this research is the result of crawling Twitter by inserting the keyword "Chat GPT”. This study method was carried out by measuring performance using the K-Nearest Neighbor and Naïve Bayes Algorithms to find the best performance value and compare it with the Text-Preprocessing generated by Rapidminer. It is shown that the performance accuracy produced using the K-Nearest Neighbor Algorithm is 73.57% using the Linear Sampling method. The comparison result with the Text-Preprocessing method using Rapidminer indeed shows a better accuracy of 75.33%, which means it has a narrow difference of 1.76% with the Chat GPT Text Pre-Processing method. However, both are still in the same category, which is Fair Classification. The results of this research show that Chat GPT can be an alternative in Text-Preprocessing datasets for sentiment analysis.

TF-TDA: A Novel Supervised Term Weighting Scheme for Sentiment Analysis

In text classification tasks, such as sentiment analysis (SA), feature representation and weighting schemes play a crucial role in classification performance. Traditional term weighting schemes depend on the term frequency within the entire document collection; therefore, they are called unsupervised term weighting (UTW) schemes. One of the most popular UTW schemes is term frequency–inverse document frequency (TF-IDF); however, this is not sufficient for SA tasks. Newer weighting schemes have been developed to take advantage of the membership of documents in their categories. These are called supervised term weighting (STW) schemes; however, most of them weigh the extracted features without considering the characteristics of some noisy features and data imbalances. Therefore, in this study, a novel STW approach was proposed, known as term frequency–term discrimination ability (TF-TDA). TF-TDA mainly presents the extracted features with different degrees of discrimination by categorizing them into several groups. Subsequently, each group is weighted based on its contribution. The proposed method was examined over four SA datasets using naive Bayes (NB) and support vector machine (SVM) models. The experimental results proved the superiority of TF-TDA over two baseline term weighting approaches, with improvements ranging from 0.52% to 3.99% in the F1 score. The statistical test results verified the significant improvement obtained by TF-TDA in most cases, where the p-value ranged from 0.0000597 to 0.0455.

RoBERTa-GRU: A Hybrid Deep Learning Model for Enhanced Sentiment Analysis

This paper proposes a novel hybrid model for sentiment analysis. The model leverages the strengths of both the Transformer model, represented by the Robustly Optimized BERT Pretraining Approach (RoBERTa), and the Recurrent Neural Network, represented by Gated Recurrent Units (GRU). The RoBERTa model provides the capability to project the texts into a discriminative embedding space through its attention mechanism, while the GRU model captures the long-range dependencies of the embedding and addresses the vanishing gradients problem. To overcome the challenge of imbalanced datasets in sentiment analysis, this paper also proposes the use of data augmentation with word embeddings by over-sampling the minority classes. This enhances the representation capacity of the model, making it more robust and accurate in handling the sentiment classification task. The proposed RoBERTa-GRU model was evaluated on three widely used sentiment analysis datasets: IMDb, Sentiment140, and Twitter US Airline Sentiment. The results show that the model achieved an accuracy of 94.63% on IMDb, 89.59% on Sentiment140, and 91.52% on Twitter US Airline Sentiment. These results demonstrate the effectiveness of the proposed RoBERTa-GRU hybrid model in sentiment analysis.

Enhancing the Generalization for Text Classification through Fusion of Backward Features.

Generalization has always been a keyword in deep learning. Pretrained models and domain adaptation technology have received widespread attention in solving the problem of generalization. They are all focused on finding features in data to improve the generalization ability and to prevent overfitting. Although they have achieved good results in various tasks, those models are unstable when classifying a sentence whose label is positive but still contains negative phrases. In this article, we analyzed the attention heat map of the benchmarks and found that previous models pay more attention to the phrase rather than to the semantic information of the whole sentence. Moreover, we proposed a method to scatter the attention away from opposite sentiment words to avoid a one-sided judgment. We designed a two-stream network and stacked the gradient reversal layer and feature projection layer within the auxiliary network. The gradient reversal layer can reverse the gradient of features in the training stage so that the parameters are optimized following the reversed gradient in the backpropagation stage. We utilized an auxiliary network to extract the backward features and then fed them into the main network to merge them with normal features extracted by the main network. We applied this method to the three baselines of TextCNN, BERT, and RoBERTa using sentiment analysis and sarcasm detection datasets. The results show that our method can improve the sentiment analysis datasets by 0.5% and the sarcasm detection datasets by 2.1%.

SART & COVIDSentiRo: Datasets for Sentiment Analysis Applied to Analyzing COVID-19 Vaccination Perception in Romanian Tweets

Vaccination is an important subject of discussion adjacent to the COVID-19 pandemic. Sentiments generated online by this topic are worth analyzing using opinion mining tools, and it is interesting to do so in online content written in an under-researched language, like Romanian. For this reason, we modified and enlarged an existing sentiment analysis dataset comprised of Romanian tweets labeled as negative or positive. The resulting dataset, SART (Sentiment Analysis from Romanian Tweets), comprised of three classes (positive, negative, and neutral) containing 1300 Romanian tweets each, was used to train two different sentiment analysis models: a fastText-based one and a fine-tuned BERT model. We further show the usefulness of the sentiment analysis model by analyzing the sentiment of Romanian tweets regarding vaccination using a corpus created and collected by the authors between January 2021 and February 2022 (COVIDSentiRo).

An Ontology-based Sentiment Analysis Approach To Discovering Hidden Affected Objects

Nowadays, the topic of sentiment analysis has attracted a wide range of artificial intelligence technologies suchas Natural Language Processing (NLP), neural network. Long Short-Term Memory (LSTM), a deep neural network, has proven the effiency in the task of sentiment classification. However, there is a shortage of semantic knowledge from such classification results. This paper introduces a sentiment analysis approach based on a knowledge model to discover hidden affected objects, which combines ONtology and SentiMent technology, named ONSEM. In order to confirm the efficiency of the ONSEM framework, an experiment on a sentiment analysis dataset was developed and then yielded promising experimental results.

Sentiment Analysis of Online Lectures using K-Nearest Neighbors based on Feature Selection

Online lecture is a distance learning system that utilizes information technology in its implementation. Although it has been agreed, this lecture system has caused controversy. Not infrequently online lectures are considered to bring a variety of new obstacles in lectures, and not a few also consider that online lectures are the most appropriate solution to continue to run lecture activities in the midst of alarming pandemic conditions. In response to this policy, many people expressed various kinds of opinions and views on the implementation of online lectures which are generally stated on social media, one of which is through Twitter. Sentiment analysis is a branch of the science of machine learning that is carried out to obtain useful information or new knowledge by extracting, understanding, and processing text data automatically. Several methods are widely used by researchers to classify sentiment analysis datasets including K-Nearest Neighbor (K-NN). K-NN will be adapted to classify online lecture datasets because K-NN can produce good accuracy on a large number of data. The presence of feature selection also helps machine learning in improving its performance. The purpose of this study was to determine student sentiment toward online lectures and to determine the level of accuracy of the combination of K-NN with various feature selections. Based on 780 tweets data, a classification of 394 positive sentiments, 320 negative sentiments, and 39 neutral sentiments was obtained. So, the results of student opinions are on POSITIVE sentiments. The accuracy result of the K-NN Algorithm was 56% and the accuracy of the K-NN Algorithm + Forward Selection was 51%, the accuracy of the KNN Algorithm + Adabost was 54%, and the accuracy of the KNN Algorithm + Genetic Algorithm was 55%.

Sentiment analysis of Indonesian reviews using fine-tuning IndoBERT and R-CNN

Reviews are a form of user experience information on a product or service that can be used as a reference for potential consumers’ preferences to buy, use, or consume a product. They can be also used by business entities to find out public opinion about their product or the performance of their business products. It will be very difficult to process the review data manually and it will take a long time. Therefore, sentiment analysis automation can be used to get polarity information from existing reviews. In this study, IndoBERT with Recurrent Convolutional Neural Network (RCNN) was used to automate sentiment analysis of Indonesian reviews. The data used was a sentiment analysis dataset obtained from IndoNLU with sentiment consisting of negative sentiment, neutral sentiment, and positive sentiment. The results of the test showed that IndoBERT with the Recurrent Convolutional Neural Network (RCNN) had better results than the IndoBERT base. IndoBERT with Recurrent Convolutional Neural Network (RCNN) obtained 95.16% accuracy, 94.05% precision, 92.74% recall and 93.27% f1 score.