Sentiment Polarity Research Articles

Beyond Sentiment Exploring the Dynamics of AIGC-Generated Sports Content and User Engagement on Xiaohongshu

This study explores the impact of AIGC-generated sports content on user sentiment and engagement on Xiaohongshu, a Chinese social media platform.Using MediaCrawler, we collected posts and comments related to sports topics generated by AI. Sentiment analysis was performed using SnowNLP to classify comments as positive or negative, with 57.1% showing positive sentiment and 42.9% negative. Despite the prevalence of positive sentiment, no significant difference in engagement (measured by likes) was observed between positive and negative comments.Temporal analysis revealed that user engagement is event-driven, with spikes in positive sentiment during major sports events. Additionally, word cloud analysis highlighted promotional and spam-like themes, suggesting challenges in content moderation. The interaction patterns showed a skewed distribution, with a small number of highly active users dominating discussions.The findings suggest that timing, relevance, and event alignment are more critical than sentiment polarity in driving engagement.The study provides insights for content creators and platforms on optimizing AIGC strategies through trend alignment and effective user collaboration.

An automated approach to health misinformation monitoring on YouTube

Abstract Background YouTube is a social media platform associated with large viewership but little research into its role in the propagation of online health misinformation. This study proposes an automated pipeline to facilitate the collection and analysis of health misinformation on YouTube. Methods The pipeline relies on Python and the Youtube Data API. A preliminary test of the proposed pipeline was conducted using two videos from the channel “@BobbyParrish” (5.55M subscribers, 1.5K videos). The pipeline was used to extract two videos with large view counts and comparable like and comment counts. This extraction includes the transcript of the respective videos and engagement metrics. All the comment threads under the videos are also collected with the reply structure preserved. Then, the pipeline employs NLTK’s SentimentIntensityAnalyzer to score each comment for sentiment polarity and classify into positive, negative, or neutral. The pipeline generates visualizations of the sentiment distribution and a frequency-based word cloud of emojis extracted from the text. Results The proposed pipeline passed the test satisfactorily. It was able to retrieve channel statistics and metrics associated with the videos on the channel. It also successfully extracted the transcript and complete comments of the videos while preserving the integrity of the reply structure found on YouTube. Automated analyses of the data resulted in comprehensive and accessible visualizations. Conclusions The proposed work has the potential to facilitate large-scale studies into the propagation of health misinformation on YouTube. Moreover, it can be used by public health officials to rapidly address viral videos spreading health misinformation through social inoculation. Future work includes integrating topic modelling and automatic classification of health misinformation in the analysis portion of the pipeline. Key messages • The pipeline offers public health officials and researchers a rapid tool to identify and analyze health misinformation on YouTube, facilitating timely interventions. • The pipeline enhances the ability to monitor and respond to evolving public health misinformation trends, by automating the extraction and sentiment analysis of YouTube data.

Do Financial Community’s Tweet Sentiments Affect the Stock Market? Evidence from an Emerging Economy

In the era of social media and internet, Twitter (presently ‘X’) has been recognized as the prominent factor for financial markets by multiple corporates and experts. The motivation of present research hinges on burgeoning popularity of Twitter in the financial community for sharing information and impact of their hidden sentiments on financial markets. Research on tweet-based sentiments and their influence on financial markets in India is still in the infancy stage. To cover this gap, the present study empirically investigates the predictive impact of financial tweet sentiments on the Indian stock Index’s returns over a short period. The paper scraped 22,816 tweets and Nifty 50 index’s intraday hour end values for January 2023. Using Natural Processing Technique, the overall sentiments of the financial community are found as highly positive for each trading hour. Next, it presents a low positive correlation between positive sentiments, sentiment polarity, and hourly returns. Lagged negative sentiments reflect the better ability to impact current period returns than positive sentiments and polarity. Additionally, sentiment polarity and returns reflect bi-directional causation for the future period impact. Investors, stock market traders, and other practitioners can employ the proven concept of linkage between tweet sentiments and returns for constructing effective investment and trading strategies. JEL Classifications: G41, D53, G17, G02, G19

Heterogeneous Hierarchical Fusion Network for Multimodal Sentiment Analysis in Real-World Environments

Multimodal sentiment analysis models can determine users’ sentiments by utilizing rich information from various sources (e.g., textual, visual, and audio). However, there are two key challenges when deploying the model in real-world environments: (1) the limitations of relying on the performance of automatic speech recognition (ASR) models can lead to errors in recognizing sentiment words, which may mislead the sentiment analysis of the textual modality, and (2) variations in information density across modalities complicate the development of a high-quality fusion framework. To address these challenges, this paper proposes a novel Multimodal Sentiment Word Optimization Module and a heterogeneous hierarchical fusion (MSWOHHF) framework. Specifically, the proposed Multimodal Sentiment Word Optimization Module optimizes the sentiment words extracted from the textual modality by the ASR model, thereby reducing sentiment word recognition errors. In the multimodal fusion phase, a heterogeneous hierarchical fusion network architecture is introduced, which first utilizes a Transformer Aggregation Module to fuse the visual and audio modalities, enhancing the high-level semantic features of each modality. A Cross-Attention Fusion Module then integrates the textual modality with the audiovisual fusion. Next, a Feature-Based Attention Fusion Module is proposed that enables fusion by dynamically tuning the weights of both the combined and unimodal representations. It then predicts sentiment polarity using a nonlinear neural network. Finally, the experimental results on the MOSI-SpeechBrain, MOSI-IBM, and MOSI-iFlytek datasets show that the MSWOHHF outperforms several baselines, demonstrating better performance.

THE POTENTIAL OF ARTIFICIAL INTELLIGENCE (AI) TO IMPROVE ELECTRONIC WORD-OF-MOUTH'S (eWOM) EFFICACY

Internet-mediated online communication, particularly with regard to a product, brand, or organization, is known as electronic word-of-mouth (eWOM). Analyzing this open exchange of opinions and information about a company or product among consumers can be extremely useful for businesses. Opinion mining, (sentiment analysis), is a popular subdomain in Natural language processing (NLP) which allows for transforming qualitative data into quantitative information. The sentiment analysis of eWOM has greatly improved with the advancement of artificial intelligence (AI). Nowadays, computer algorithms can automatically classify the sentiment polarity of digital communication after extracting plain text. Artificial intelligence (AI) has the potential to fundamentally alter how companies assess and use consumer feedback to improve their products and services. In this paper, the authors, by analysing the attitudes of 450 respondents, tried to bring this current topic closer to experts in the field of digital marketing, in order to point out to them all the benefits that the sentiment analysis of consumers with the help of artificial intelligence algorithms can provide. The aim of this study is to indicate that if marketing experts use sentiment analysis supported by artificial intelligence (AI), they will be able to gain deeper insights on their customers and adjust their business strategies accordingly.

Hierarchical graph contrastive learning framework based on quantum neural networks for sentiment analysis

Existing multi-modal sentiment analysis (MSA) methods typically achieve interaction by connecting different layers or designing special structures, but rarely consider the synergistic effects among data. Moreover, most sentiment analysis research tends to focus solely on single sentiment polarity analysis, without considering the intensity and directional attributes of emotions. Addressing these issues, we propose a framework called Hierarchical Graph Contrastive Learning based on Quantum Neural Network (HGCL-QNN) to remedy these shortcomings. Specifically, a graph structure is established within and between modalities. In the quantum fuzzy neural network module, fuzzy quantum encoding is implemented by using complex-valued, then quantum superposition and entanglement are utilized to consider the intensity and directional attributes of emotions while analyzing emotional polarity. In the quantum multi-modal fusion neural network module, methods such as amplitude encoding and quantum entanglement are employed to further integrate information from different modalities, thereby enhancing the model's power to express emotional information. To enhance the model's understanding of fine-grained and global features, and to better align and integrate features from different modalities, hierarchical graph contrastive learning is employed on different levels. The experimental results demonstrate that HGCL-QNN outperforms the existing baseline methods on MOSI and MOSEI datasets, achieving significant efficacy improvements.

Dual syntax aware graph attention networks with prompt for aspect-based sentiment analysis

Aspect-based sentiment analysis (ABSA) is a challenging task due to the presence of multiple aspect words with different sentiment polarities in a sentence. Recently, pre-trained language models like BERT have been widely used as context encoders in ABSA. Graph neural networks have also been employed to extract syntactic and semantic information from sentence parsing trees, resulting in superior results. However, dependency trees may establish irrelevant dependencies for sentences with irregular syntax and complex structures. Additionally, previous methods have not fully utilized recent developments in pre-trained language models. Therefore, we propose a Dual Syntax aware Graph attention networks with Prompt (DSGP) model to address these issues. Our model utilizes prompt templates to maximize the potential of pre-trained models and masked vector outputs of templates as supplementary aspect feature representations. We also leverage both dependency trees and constituent trees with graph attention networks to extract different types of syntactic information. The dependency tree captures syntactic correlation between words, while the constituent tree provides a high-level formation of the sentence. Finally, the output from the prompt and parsing trees is fused and fed into a standard classifier. Experimental results on four public datasets demonstrate the competitive performance of our model.

"News has various shades": Quantifying and Analyzing Media Bias at Aspect-level Granularity

This research presents a novel approach to dissecting media narratives, with a specific focus on aspect-level granularity and variance in addressing multifaceted news topics. Unlike previous studies, which often concentrate solely on ideological or political biases, our methodology delves deeper, exploring how diverse media outlets navigate the complexities of various news aspects. Through a detailed case study on Indian government policies, we uncover distinctive biases and variations in reporting between Indian and international media. Crucially, our methodology leverages Natural Language Inference (NLI) to identify news aspects and ascertain aspect-level sentiment from news text, enabling scalable and precise quantification of bias across diverse media narratives. Our findings illuminate the multifaceted layers of media coverage, revealing nuanced stances on different aspects of the same topic and the dynamic nature of biases over time. Importantly, our comprehensive framework quantifies media bias through data-driven quantitative evaluation, capturing both selection bias based on news aspect coverage and statement bias based on sentiment polarity. It is relevant to note that in our context, bias is not inherently favourable or unfavourable but rather serves as a quantitative metric to measure divergence in news presentation from the overall average. Thus, our contributions not only enhance understanding of media bias but also introduce a methodology distinguished for its comprehensive and scalable approach to analyzing media narratives, with substantial implications for future research and discussions in this domain.

Sentiment analysis in medication adherence: using ruled-based and artificial intelligence-driven algorithms to understand patient medication experiences.

Studies are exploring ways to improve medication adherence, with sentiment analysis (SA) being an underutilized innovation in pharmacy. This technique uses artificial intelligence (AI) and natural language processing to assess text for underlying feelings and emotions. This study aimed to evaluate the use of two SA models, Valence Aware Dictionary for Sentiment Reasoning (VADER) and Emotion English DistilRoBERTa-base (DistilRoBERTa), for the identification of patients' sentiments and emotions towards their pharmacotherapy. A dataset containing 320,095 anonymized patients' reports of experiences with their medication was used. VADER assessed sentiment polarity on a scale from - 1 (negative) to + 1 (positive). DistilRoBERTa classified emotions into seven categories: anger, disgust, fear, joy, neutral, sadness, and surprise. Performance metrics for the models were obtained using the sklearn.metrics module of scikit-learn in Python. VADER demonstrated an overall accuracy of 0.70. For negative sentiments, it achieved a precision of 0.68, recall of 0.80, and an F1-score of 0.73, while for positive sentiments, it had a precision of 0.73, recall of 0.59, and an F1-score of 0.65. The AUC for the ROC curve was 0.90. DistilRoBERTa analysis showed that higher ratings for medication effectiveness, ease of use, and satisfaction corresponded with more positive emotional responses. These results were consistent with VADER's sentiment analysis, confirming the reliability of both models. VADER and DistilRoBERTa effectively analyzed patients' sentiments towards pharmacotherapy, providing valuable information. These findings encourage studies of SA in clinical pharmacy practice, paving the way for more personalized and effective patient care strategies.

Dual edge-embedding graph convolutional network for unified aspect-based sentiment analysis

Aspect-based sentiment analysis (ABSA) mainly involves three elements, i.e., aspect, target, and sentiment. Most existing research focuses on aspect-sentiment or target-sentiment separately, while connection among the three elements is still underexploited. In order to extract three interrelated elements from a sentence simultaneously, we propose a unified end-to-end model, which is based on one aspect via a virtual edge attention mechanism and a graph convolutional network (GCN) to detect related targets and to analyze the corresponding sentiment polarity. The virtual edges are decided by the association intensity between the aspect and the words of the sentence. Instead of naively regarding all connecting edges as binary values, we design a novel GCN with a dual edge-embedding method to help information flow among each word in the sentence and between the aspect category and the sentence. Extensive experiments on two open datasets show that our model outperforms the state-of-the-art models in the aspect-target-sentiment triples detection task and its two subtasks.

Implicit expression recognition enhanced table-filling for aspect sentiment triplet extraction

Aspect sentiment triplet extraction (ASTE) is a challenging task in aspect-based sentiment analysis (ABSA), involving the identification of aspect terms, opinion terms, and their corresponding sentiment polarities within comments to form triplets. The emergence of more realistic DMASTE datasets, featuring diverse domains, implicit aspect terms, and longer comments, poses challenges for existing methods. In particular, these methods struggle with recognizing implicit expressions effectively and capturing sufficient information. To overcome these hurdles, we propose an implicit expression recognition enhanced table-filling (IERET) method. This approach integrates modeling of overall implicit expression in sentences and employs a bidirectional information aggregation module to capture word pair information comprehensively. During the decoding process, a table-filling method accurately delineates aspect-opinion pair boundaries. Experimental results across in-domain, single-source cross-domain, and multi-source cross-domain on the DMASTE dataset demonstrate that our proposed IERET method achieves state-of-the-art performance.

Aspect Based Feature Extraction in Sentiment Analysis using Bi-GRU-LSTM Model

In Natural Language Processing (NLP), Sentiment Analysis (SA) is a fundamental process which predicts the sentiment expressed in sentences. In contrast to conventional sentiment analysis, Aspect-Based Sentiment Analysis (ABSA) employs a more nuanced approach to assess the sentiment of individual aspects or components within a document or sentence. Its objective is to identify the sentiment polarity, such as positive, neutral, or negative, associated with particular elements disclosed within a sentence. This research introduces a novel sentiment analysis technique that proves to be more efficient in sentiment analysis compared to current methods. The suggested sentiment analysis method undergoes three key phases: 1. Pre-processing 2. Extraction of aspect sentiment and 3. Sentiment analysis classification. The input text data undergoes pre-processing through the implementation of four typical text normalization techniques, which include stemming, stop word elimination, lemmatization, and tokenization. By employing these methods, the provided text data is prepared and fed into the aspect sentiment extraction phase. During the aspect sentiment extraction phase, features are obtained through a series of steps, including enhanced ATE (Aspect Term Extraction), assessment of word length, and determination of cosine similarity. By following these steps, the relevant features are extracted on the basis of aspects and sentiments involved in the text data. Further, a hybrid classification model is proposed to classify sentiments. In this work, two of the Deep Learning (DL) classifiers, Bi-directional Gated Recurrent Unit (Bi-GRU) and Long Short-Term memory (LSTM) are used in proposing a hybrid classification model which classifies the sentiments effectively and provides accurate final predicted results. Moreover, the performance of proposed sentiment analysis technique is analyzed experimentally to show its efficacy over other models.

IDSV-GCN: Integrating Dual Syntactic Views Graph Convolutional Network for aspect-based sentiment analysis

Aspect-based sentiment analysis (ABSA) aims to predict the sentiment polarity towards given aspects in a sentence. Recent research has highlighted that graph neural networks based on syntactic dependency trees effectively convey rich syntactic information and are suitable for ABSA tasks. Most existing work typically adopt either dependency type or minimum distance between nodes in the original syntactic tree to construct different views. However, existing work overlooks the high-dimensional local feature information in dependency relations, resulting in inadequate exploration of this view. Meanwhile, most work ignores that the minimum distance view and the dependency view can be complementary and integrated in syntactic analysis. Unfortunately, directly adding the two views introduces noise issues, leading to unsatisfactory results after their integration. To address these issues, this paper introduces an integrating dual syntactic views graph convolutional network (IDSV-GCN) model that seamlessly integrates dependency type view and minimum distance view, enabling comprehensive utilization of syntactic structure and information. To better address the computation of weights in the dependency type view, we innovatively propose a fusion attention mechanism. This approach effectively captures both global and local feature information. To more effectively integrate the two views and reduce the impact of noise, we employ a syntactic mask layer to filter out noise in the dependency type view and highlighting important dependency feature information. Then, we construct a minimum distance weight matrix based on the matrix obtained from the dependency type view. Finally, we merge the two matrices and feed them into the GCN to facilitate the model’s learning of syntactic information. Experimental results on three widely-used benchmark datasets demonstrate that our proposed IDSV-GCN model outperforms state-of-the-art methods.

Analyzing World Cup Impact Through an Evolutionary Optimization Approach Based on Sentiment Polarity with Pre-trained Word Embeddings

Analyzing World Cup Impact Through an Evolutionary Optimization Approach Based on Sentiment Polarity with Pre-trained Word Embeddings

A survey on pragmatic processing techniques

Pragmatics, situated in the domains of linguistics and computational linguistics, explores the influence of context on language interpretation, extending beyond the literal meaning of expressions. It constitutes a fundamental element for natural language understanding in machine intelligence. With the advancement of large language models, the research focus in natural language processing has predominantly shifted toward high-level task processing, inadvertently downplaying the importance of foundational pragmatic processing tasks. Nevertheless, pragmatics serves as a crucial medium for unraveling human language cognition. The exploration of pragmatic processing stands as a pivotal facet in realizing linguistic intelligence. This survey encompasses important pragmatic processing techniques for subjective and emotive tasks, such as personality recognition, sarcasm detection, metaphor understanding, aspect extraction, and sentiment polarity detection. It spans theoretical research, the forefront of pragmatic processing techniques, and downstream applications, aiming to highlight the significance of these low-level tasks in advancing natural language understanding and linguistic intelligence.

Knowledge complementation based dual-table filling approach for aspect sentiment triplet extraction

Aspect Sentiment Triplet Extraction (ASTE) task aims to extract aspect terms, opinion terms, and the corresponding sentiment polarity of aspect-opinion pairs from review sentences. Current outperforming table-filling approaches focus on enhancing the relation representation within a single table, with various algorithms designed to exploit deeper information and additional knowledge. However, due to the sequential execution of these algorithms on a single table, the granularity of the knowledge evolves unidirectionally, inevitably leading to the loss of knowledge at different granularities from the previous stage of the entire process. To overcome this limitation, we propose a simple yet effective end-to-end model for the ASTE task, namely Dual-Table Filling (DTF), which constructs two complementary tables instead of just one to capture different granularities of knowledge in a sentence. Specifically, the main table exploits latent knowledge and hidden dependencies, while the assistant table captures more fine-grained contextual knowledge. This structure allows the model to utilize knowledge at multiple granularities more efficiently while maintaining simplicity. In addition, a Syntactic Enhanced Encoder (SEE) is developed to acquire syntactic-enhanced knowledge from sentences, enhancing the comprehension of sentence structure and word relationships. Experimental results on public benchmark datasets show the effectiveness and exceptional performance of our proposed model.

Use of AI Applications to Learn the Sentiment Polarity of Public Perceptions: A Case Study of the COVID-19 Vaccinations in the UAE

Use of AI Applications to Learn the Sentiment Polarity of Public Perceptions: A Case Study of the COVID-19 Vaccinations in the UAE

Leveraging Large Language Models for Aspect-Based Sentiment Classification using GPT-4

Aspect-based sentiment analysis (ABSA) is an essential task in natural language processing (NLP) that aims to extract both aspects and their corresponding sentiment polarities from textual data. Despite its importance, ABSA faces challenges in accurately capturing nuanced sentiments, particularly in complex and context-dependent scenarios. This paper introduces a novel technique that employs few-shot learning with the GPT-4 model to determine the sentiment of specific aspects within the text. By using a template prompt along with few-shot examples, we enhance ChatGPT’s in-context learning capability, thereby improving the effectiveness of aspect-based sentiment classification. Our tests and evaluations demonstrate that this method achieves state-of-the-art results in ABSA. This research represents a significant advancement in sentiment analysis methodologies, especially in capturing nuanced sentiments toward specific aspects.

From Optimism to Concern: Unveiling Sentiments and Perceptions Surrounding ChatGPT on Twitter

Artificial intelligence (AI) technologies, as a product of processes aimed at imitating human intelligence through computers and software, affect our daily lives in various aspects, including cultural, technological, and economic. The advent of ChatGPT, developed by OpenAI, signifies a pivotal AI advancement in human language comprehension and generation, heralding profound implications across societal, economic, and cultural dimensions. However, a notable gap exists in scholarly literature concerning the examination of perceptions and discussions surrounding ChatGPT. This study aims to address this gap by analyzing Twitter conversations, comprising 1.1 million tweets containing “ChatGPT” or “#ChatGPT” between December 1 2022 and June 1 2023, with geolocation data. Employing a text as data approach encompassing text, sentiment, and semantic network analyses, sentiment polarity and lexical patterns were explained, while semantic network analysis revealed central expressions and prominent themes in Twitter discussions. The study highlights the social effects of AI technologies from the perspective of Twitter users and reveals the sentimental tendency and themes in tweets with geographical and economic dimensions. The findings reflect the prevalence of positive sentiment and hype toward ChatGPT while there are also concerns regarding privacy, cybersecurity, and misuse of AI tools. More importantly, content generation, technical aspects, business applications, educational use, and competitors are among the main themes of ChatGPT-related discussions on Twitter.

Fusion Network for Aspect-Level Sentiment Classification Based on Graph Neural Networks—Enhanced Syntactics and Semantics

Aspect-level sentiment classification (ALSC) struggles with correctly trapping the aspects and corresponding sentiment polarity of a statement. Recently, several works have combined the syntactic structure and semantic information of sentences for more efficient analysis. The combination of sentence knowledge with graph neural networks has also proven effective at ALSC. However, there are still limitations on how to effectively fuse syntactic structure and semantic information when dealing with complex sentence structures and informal expressions. To deal with these problems, we propose an ALSC fusion network that combines graph neural networks with a simultaneous consideration of syntactic structure and semantic information. Specifically, our model is composed of a syntactic attention module and a semantic enhancement module. First, the syntactic attention module builds a dependency parse tree with the aspect term being the root, so that the model focuses better on the words closely related to the aspect terms, and captures the syntactic structure through a graph attention network. In addition, the semantic enhancement module generates the adjacency matrix through self-attention, which is processed by the graph convolutional network to obtain the semantic details. Lastly, the extracted features are merged to achieve sentiment classification. As verified by experiments, the model we propose can effectively enhance ALSC’s behavior.