Narrow Results

Given the volatility and complexity of financial markets, accurate risk prediction is paramount for effective risk management. Traditional methods like Support Vector Machines (SVM) and Back Propagation Neural Networks (BPNNs) have shown limitations, particularly in capturing the dependencies in sequential data inherent to financial time series. This paper proposes a novel approach to financial risk prediction using Conditional Random Fields (CRFs), leveraging their potential in sequence data modeling to address these challenges. We first elaborate on the construction of a comprehensive feature set, incorporating market data, macroeconomic indicators, and sentiment analysis from financial news, tailored to enhance the predictive power of the CRFs model. The model itself is customized with carefully designed state and observation sequences, along with feature functions specifically engineered for financial risk prediction. Experiments conducted across multiple financial datasets demonstrate the superiority of our CRF-based model over traditional and machine learning benchmarks in terms of accuracy and stability. Our findings not only contribute to the literature by filling the gap left by traditional financial risk prediction methods but also offer a robust tool for practitioners in financial risk management. This work exemplifies the potential of applying advanced sequence modeling techniques to financial risk prediction, setting a new benchmark for future research in the field.

In the evolving e-commerce landscape, numerous subjective evaluations of products are surfacing on diverse online platforms. These user-generated reviews encapsulate sentiments towards various product attributes, aiding sellers in identifying the strengths and weaknesses of their offerings. Consequently, this feedback facilitates product enhancement and informs prospective buyers, enabling more judged purchasing decisions. Leveraging sentiment analysis on voluminous review datasets can elucidate customer attitudes towards specific goods or services, thus providing businesses with insights into customer emotions. This study introduces a hybrid Convolutional Neural Network combined with Bidirectional Long Short-Term Memory (CNN–BiLSTM) model designed for sentiment analysis of consumer reviews. This approach exhibits robust scalability, sustaining high-performance metrics such as accuracy, recall and F1 scores above 92%, even with the addition of fresh datasets. The proposed CNN–BiLSTM model surpasses conventional CNN and RNN models by integrating the superior predictive power of the BiLSTM component, which is adept at capturing long-range dependencies and context. Contrasted with a standalone BiLSTM model, our proposed architecture leverages the strengths of both technologies: the CNN layer for efficient feature dimensionality reduction and the BiLSTM layer for extracting temporal and contextual information. This synergy enhances the extraction of pertinent features from online consumer reviews, thereby boosting prediction accuracy and operational efficiency.

Sentiment analysis (SA) is an essential application of machine learning (ML) and natural language processing (NLP) that comprises the automatic extraction of opinions or sentiments presented in textual data. By leveraging methods to distinguish the expressive nature conveyed in written content, SA permits businesses and research workers to gain valuable insights into social media discourse, customer feedback, and public reviews. In the field of SA, the synergy of Applied Linguistics and Artificial Intelligence (AI) has led to a robust method that goes beyond conventional methods. By incorporating linguistic principles into AI methods, this interdisciplinary collaboration allows a more nuanced perception of human sentiments expressed in language. Applied Linguistics offers the theoretical basis for understanding the details of pragmatics, semantics, and linguistic structures, while AI algorithms leverage this knowledge for analyzing large datasets with notable accuracy. This study presents an Applied Linguistics-driven Artificial intelligence Approach for SA and Classification (ALAIA-SAC) system in social media. The primary intention of the ALAIA-SAC technique is to apply an attention mechanism with a fractal hyperparameter-tuned deep learning (DL) method for identifying sentiments. In the ALAIA-SAC technique, data preprocessing takes place in several stages to convert the input data into a compatible format. In addition, the TF-IDF model could be employed for the word embedding method. The self-attention directional long short-term memory (SBiLSTM) model is used for sentiment classification. Finally, the SBiLSTM model’s hyperparameter selection is performed using a Fractal Pelican optimization algorithm (FPOA). The experimentation results of the ALAIA-SAC method are assessed under two benchmark datasets. The comparative study of the ALAIA-SAC technique exhibited a superior accuracy value of 99.17% and 99.39% under Twitter US Airlines and IMDB datasets.

Sentimental analysis is one of the most complicated tasks in text mining. It is the process of converting unstructured text to structural text to identify the meaningful pattern of data. Different kinds of sentiments in social media posts are classified into positive and negative. Natural Language Processing (NLP) plays an imperative task in examining sentiments on social media. Sentimental analysis techniques frequently use text data to help to monitor customer feedback about products and understand the needs of customers in the field of business. The text data are gathered from social media comments that lead to several drawbacks such as spelling errors, noise, and unstructured data, and some of the data are in abbreviation form in which abbreviations are hard to understand from the text data. Therefore, the innovative sentiment analysis method is implemented to identify the kinds of sentiments from the collected social media posts. At first, the text data are assembled from the social media posts. The collected text data are subjected to text preprocessing techniques to improve the data quality. NLP techniques such as a bag of n-grams, glove embedding, and Bidirectional Encoder Representations from Transformers (BERT) are applied to separate the features from the text data. The removed features from these techniques are given to the fusion of the optimal feature stage, where the weight optimization takes place via the Mutated Random Parameter-based Serval Optimization Algorithm (MRP-SOA). The attained optimal weighted fused features contain the most relevant information, and it is fed to the classifier for analyzing the sentiments. For analyzing the sentiments, the Cascaded Adaptive Dilated Temporal Convolutional Network (CADTCN) is utilized, and the parameters from ADTCN are optimized with the same to get higher classification outcomes. The analysis outcomes are compared with the traditional sentiment analysis models in terms of various metrics to show the success of the advanced design.

Bank risk management is a crucial issue in the stability of the financial system. How to select high-risk factors that make banks in trouble and how these factors affect bank risks have always been a core problem. Previous studies comprehensively identified bank risk factors from textual risk disclosures and used the disclosure frequency of risk factors to determine important factors to which banks should pay more attention. This paper creatively constructs the textual risk matrix with frequency and sentiment of risk factors to divide bank risk factors into the high-risk category, mid-risk category, and low-risk category. Then we explore the impact of different categories of risk factors on bank risk and the risk perception of investors. Based on 457,383 sentences of 2,735 Form 10-K reports of 240 American commercial banks from 2006 to 2020, 33 bank risk factors were identified. Three risk factors belong to in high-risk category, including loan loss risk, regulation risk, and interest rate risk. Three factors are classified in the mid-risk category and 27 risk factors are low-risk factors. The regression results show that compared with individual bankruptcy risk, risk factors have better prediction and interpretive ability on the systemic risk. The disclosure of bank risk factors will affect the investors’ risk perception, especially the worse risk situation of the high-risk factors will increase the risk perceived by investors.

Corporate reputation is one of the most valuable assets a company must maintain to remain in business, and it becomes even more critical during crises such as COVID-19, which pose a severe threat not only to employees, customers, and the general public but also to the company’s fundamental survival. The purpose of this study is to identify themes affecting corporate reputation and assess how leading companies responded to the COVID-19 crisis. This study targeted the top 100 (RQ score higher than 50) companies according to the 23rd Annual Reputation Quotient from Axios-Harries Poll 100 RQ report. Employing cluster and sentiment analysis, the study explores CEO letters to understand how they emphasize reputation factors and employ impression management tactics. Findings of our study illustrate nuanced reactions and attitudes of CEOs toward issues presented by COVID-19, shed light on the intricacies of reputation management during times of crisis.

Effectively combating the 2019 coronavirus disease (COVID-19), pandemic relies not only on robust governmental policies but also on the active cooperation of the populace. In order to formulate rational and effective preventive policies, this study proposes an interval 2-tuple linguistic matrix game method and defines an interval 2-tuple linguistic scoring function. Building upon this framework, the study employs sentiment analysis based on the BosonNLP sentiment dictionary to extract and calculate textual scores from user comments on COVID-19 pandemic-related topics in Weibo. These two approaches are then integrated and applied to policy decision-making for the COVID-19 situation in the Shanghai region. Finally, the matrix game method is used to find the optimal prevention and control strategy in Shanghai during the epidemic period. The first is “Implement citywide preventive measures”, the second is “Implement traffic control combined with regional control policy”, and the last is “Implement the home quarantine policy”.

The Sanskrit language holds significant importance in Indian culture because it has been extensively used in religious literature, primarily in Hinduism. Numerous ancient Hindu texts originally composed in Sanskrit have since been translated into various Indian and non-Indian languages by Indian and foreign authors. These translations offer a renewed cultural perspective and broaden the reach of Indian literature to a global audience. However, the manual translations of these religious texts often lack thorough validation. Recent advancements in semantic and sentiment analysis, powered by deep learning, have provided enhanced tools for understanding language and text. In this paper, we present a framework that uses semantic and sentiment analysis to validate the English translation of the Ramayana against its original Sanskrit version. The “Ramayana” which narrates the journey of the Rama, the king of Ayodhya, is an ancient Hindu epic written by the sage Valmiki. It is known for its contribution to human values for centuries and has universal relevance. Given the importance of Sanskrit in Indian culture and its influence on literature, understanding the translations of key texts like the Ramayana is essential. Multilingual Bidirectional Encoder Representations from Transformers (mBERT) model is utilized to analyze the selected chapters of the English and the Sanskrit versions of Ramayana. Our analysis reveals that sentiment and semantic alignment between the original Sanskrit and English translations remain consistent despite stylistic and vocabulary differences. The study also compares the findings of Bidirectional Encoder Representations from Transformers (BERT) with its other variants to examine which BERT variant is more suitable for validating Sanskrit text. The paper demonstrates the potential of deep learning techniques for cross-lingual validation of ancient texts.

It is of great significance for individuals, enterprises, and government departments to analyze and excavate the sentiment in the comments. Many deep learning models are used for text sentiment analysis, and the BiTCN model has good efficacy on sentiment analysis. However, in the actual semantic expression, the contribution of each word to the sentiment tendency is different, BiTCN treats it fairly and does not pay more attention to the key sentiment words. For this problem, a sentiment analysis model based on the BiTCN-Attention is proposed in this paper. The Self-Attention mechanism and Multi-Head Self-Attention mechanism are added to BiTCN respectively to form BiTCN-SA and BiTCN-MHSA, which improve the weight of sentiment words and the accuracy of feature extraction, to increase the effect of sentiment analysis. The experimental results show that the model accuracies of BiTCN-SA and BiTCN-MHSA in the JingDong commodity review data set are 3.96% and 2.41% higher than that of BiTCN, respectively. In the comment data set of DianPing, the accuracy of BiTCN-SA and BiTCN-MHSA improved by 4.62% and 3.49%, respectively, compared with that of BiTCN.

Automated sentiment analysis is becoming increasingly recognized due to the growing importance of social media and e-commerce platform review websites. Deep neural networks outperform traditional lexicon-based and machine learning methods by effectively exploiting contextual word embeddings to generate dense document representation. However, this representation model is not fully adequate to capture topical semantics and the sentiment polarity of words. To overcome these problems, a novel sentiment analysis model is proposed that utilizes richer document representations of word-emotion associations and topic models, which is the main computational novelty of this study. The sentiment analysis model integrates word embeddings with lexicon-based sentiment and emotion indicators, including negations and emoticons, and to further improve its performance, a topic modeling component is utilized together with a bag-of-words model based on a supervised term weighting scheme. The effectiveness of the proposed model is evaluated using large datasets of Amazon product reviews and hotel reviews. Experimental results prove that the proposed document representation is valid for the sentiment analysis of product and hotel reviews, irrespective of their class imbalance. The results also show that the proposed model improves on existing machine learning methods.

Considering the 2030 United Nations intent of world connection, Cyber Intelligence becomes the main area of the human dimension able of inflicting changes in geopolitical dynamics. In cyberspace, the new battlefield is the mind of people including new weapons like abuse of social media with information manipulation, deception by activists and misinformation. In this paper, a Sentiment Analysis system with Anomaly Detection (SAAD) capability is proposed. The system, scalable and modular, uses an OSINT-Deep Learning approach to investigate on social media sentiment in order to predict suspicious anomaly trend in Twitter posts. Anomaly detection is investigated with a new semi-supervised process that is able to detect potentially dangerous situations in critical areas. The main contributions of the paper are the system suitability for working in different areas and domains, the anomaly detection procedure in sentiment context and a time-dependent confusion matrix to address model evaluation with unbalanced dataset. Real experiments and tests were performed on Sahel Region. The detected anomalies in negative sentiment have been checked by experts of Sahel area, proving true links between the models results and real situations observable from the tweets.

LSTM-SNP model is a recently developed long short-term memory (LSTM) network, which is inspired from the mechanisms of spiking neural P (SNP) systems. In this paper, LSTM-SNP is utilized to propose a novel model for aspect-level sentiment analysis, termed as ALS model. The LSTM-SNP model has three gates: reset gate, consumption gate and generation gate. Moreover, attention mechanism is integrated with LSTM-SNP model. The ALS model can better capture the sentiment features in the text to compute the correlation between context and aspect words. To validate the effectiveness of the ALS model for aspect-level sentiment analysis, comparison experiments with 17 baseline models are conducted on three real-life data sets. The experimental results demonstrate that the ALS model has a simpler structure and can achieve better performance compared to these baseline models.

With the rapid development of the Internet and big data technologies, a rich of online data (including news releases) can helpfully facilitate forecasting oil price trends. Accordingly, this study introduces sentiment analysis, a useful big data analysis tool, to understand the relevant information of online news articles and formulate an oil price trend prediction method with sentiment. Three main steps are included in the proposed method, i.e., sentiment analysis, relationship investigation and trend prediction. In sentiment analysis, the sentiment (or tone) is extracted based on a dictionary-based approach to capture the relevant online information concerning oil markets and the driving factors. In relationship investigation, the Granger causality analysis is conducted to explore whether and how the sentiment impacts oil price. In trend prediction, the sentiment is used as an important independent variable, and some popular forecasting models, e.g., logistic regression, support vector machine, decision tree and back propagation neural network, are performed. With crude oil futures prices of the West Texas Intermediate (WTI) and news articles of the Thomson Reuters as studying samples, the empirical results statistically support the powerful predictive power of sentiment for oil price trends and hence the effectiveness of the proposed method.

Link prediction is an important topic in the complex networks which aims to mine the missed or unobserved links and find the possible connected links in the network according to the known network structure. In this paper, we proposed a novel link prediction method for the directed graph which considers link semantic. The experimental results on the wiki vote network show that our link prediction method integrating link attributes achieved better results than the other compared methods.

This paper proposes a novel hybrid embedding to enhance scope of word embeddings by augmenting these with natural language processing operations. We primarily focus on the proposal of new hybrid word embedding generated by augmenting BERT embedding vectors with polarity score. The paper further proposes a new deep learning architecture inspired by the use of convolutional neural network for feature extraction and a bidirectional recurrent network for contextual and temporal feature exploitation. Use of CNN with hybrid embedding allowed the network to extract even the higher-level styles in writing, while bidirectional RNN helped in understanding context. The paper justifies that the proposed architecture and hybrid embedding improves performance of sentiment classification system by performing a large number of experiments and testing on a number of deep learning architectures. The architecture on new hybrid embeddings incurred an accuracy of 96%, which is a significant improvement when compared with recent studies in the literature.

The response to a natural disaster ultimately depends on credible and real-time information regarding impacted people and areas. Nowadays, social media platforms such as Twitter have emerged as the primary and fastest means of disseminating information. Due to the massive, imprecise, and redundant information on Twitter, efficient automatic sentiment analysis (SA) plays a crucial role in enhancing disaster response. This paper proposes a novel methodology to efficiently perform SA of Twitter data during a natural disaster. The tweets during a natural calamity are biased toward the negative polarity, producing imbalanced data. The proposed methodology has reduced the misclassification of minority class samples through the adaptive synthetic sampling technique. A binary modified equilibrium optimizer has been used to remove irrelevant and redundant features. The k-nearest neighbor has been used for sentiment classification with the optimized value of k. The nine datasets on natural disasters have been used for evaluation. The performance of the proposed methodology has been validated using the Friedman mean rank test against nine state-of-the-art techniques, including two optimized, one transfer learning, one deep learning, two ensemble learning, and three baseline classifiers. The results show the significance of the proposed methodology through the average improvement of 6.9%, 13.3%, 20.2%, and 18% for accuracy, precision, recall, and F1-score, respectively, as compared to nine state-of-the-art techniques.

Sentiment analysis using scene text images is complex and challenging because it has an arbitrary background, and the method should rely on only visual features. Unlike most existing methods that use either text or images or both, this study uses only scene text images for sentiment analysis. The intuition to use only scene text images is that sometimes users express their feelings and emotions or convey their messages by writing text in different shapes with diverse background designs. It is noted that the existing methods ignore such vital cues for sentiment analysis. This work explores a vision transformer to extract visual features that represent contextual information about the appearance of the text image. Further, to strengthen the visual features, the proposed work introduces contrastive learning which maximizes the gap between inter-classes and minimizes the gap between intra-classes of positive, negative, and neutral. To demonstrate the effectiveness of the proposed method, it is tested on our own constructed dataset and benchmark dataset. A comparative study of our method with the existing method shows the proposed method is superior in the classification of positive, negative, and neutral scene text images.

Aiming at the problem of low accuracy rate of current sentiment analysis methods for book review texts, a book review sentiment analysis method based on BERT-ABiLSTM hybrid model is proposed. First, the overall framework of sentiment analysis is constructed by integrating sentiment vocabulary and deep learning methods, and the fine-grained sentiment analysis is divided into three stages: topic identification, sentiment identification and thematic sentiment identification. Then, a dynamic character-level word vector containing contextual information is generated using a bidirectional encoder representation from transformers (BERT) pre-trained language model. Then, the contextual information in the text data is fully learned by introducing the bidirectional long short-term memory (BiLSTM) model. Finally, the accurate analysis of book review sentiment is achieved by using Attention mechanism to highlight important features and improve the efficiency of resource utilization. Through an experimental comparison with existing advanced algorithms, the proposed method in this study has improved at least 4.2%, 3.9% and 3.79% in precision, recall and F1 values, respectively. The experimental results show that the proposed BERT-ABiLSTM is higher than the existing models under different metrics, indicating that the proposed model has a good application prospect in the fields of book review analysis and book recommendation.

The evaluation of feedback collected from students at the end of the year is very essential for every educational institution. It is important to improve the teaching–learning process and the annual appraisal process. The existing approach utilizes a Likert scale questionnaire, which allows students to express their level of agreement or disagreement with given statements or provide a neutral response. Additionally, the feedback form includes open-ended questions where students can provide textual feedback. This study introduces a Lexicon-based approach to automatically analyze the textual feedback concerning different aspects of teaching. Aspect-based Sentiment Analysis (ABSA) of student feedback aims to identify sentiments expressed toward various aspects of teachers, such as their ability to address student doubts and their overall knowledge. This study explores linguistic characteristics found in sentences, including negation, modifiers and contact shifters. To assess the sentiment of a sentence, the SentiWordNet lexicon is utilized to assign scores to individual words. Based on these scores, the sentence is categorized as either positive, negative or neutral. According to the experimental findings, the Aspect-Oriented Lexicon-Based (AOLB) approach demonstrates superior performance compared to other baseline methods when it comes to accurately scoring sentiment. The approach achieved a high accuracy rate of 94% for the student feedback dataset-I, 74% for the student feedback dataset-II, 55% for laptop reviews and 59% for restaurant reviews in the SemEval 2014 dataset-III.

The monitoring and early warning of financial risks have become a crucial link in maintaining market stability and safeguarding the rights and interests of investors. Traditional financial risk monitoring methods often rely on a single data source or analysis model, making it challenging to comprehensively and accurately capture risk signals. Therefore, this paper proposes a novel financial risk monitoring model based on multimodal neural networks, which innovatively integrates multiple data sources, such as vision, language and audio, and utilizes their inherent correlations to enhance the accuracy of risk identification. First, by employing the Bidirectional Long Short-Term Memory Network (BiLSTM) structure and incorporating the self-attention mechanism, the semantic information of financial texts is deeply analyzed through the calculation of dynamic weight coefficients. Additionally, Option-based Hierarchical Reinforcement Learning (OHRL) is utilized to accurately model the behavior of market participants, capturing nuanced changes in their decision-making process. By integrating these two types of information, a comprehensive BiLSTM-OHRL model is formulated to evaluate the risk status of financial markets in a more comprehensive and accurate manner. The results demonstrate that the model performs impressively in financial risk monitoring, accurately capturing the emotional and behavioral characteristics of market participants, thereby enhancing the comprehensiveness and predictive capability of the monitoring model. It provides robust technical support for the stable operation of the financial market.