Text Databases Research Articles

A systematic review and meta-analysis of the effectiveness of social support on turnover intention in clinical nurses.

Nurse turnover has become a salient issue in healthcare system worldwide and seriously compromises patient outcomes. Social support is considered an effective contributor to alleviate nurse turnover intention (TI). However, the degree of correlation between social support and nurse TI remains elusive. This study aims to evaluate the strength of the effectiveness of social support on TI among nurses as well as its potential moderators. This systematic review and meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses. To obtained qualified studies, two researchers searched Embase, PubMed, Web of science, CINAHL, CNKI, WanFang, and Chinese Medical Journal Full Text Database from inception to January 6, 2024. Meta-analysis, publication bias, and sensitivity analysis were carried out on the included studies using CMA 3.0 software, and the moderating effect was verified through meta-analysis of variance (ANOVA). A total of 38 studies were obtained, involving 63,989 clinical nurses. The comprehensive effect size of the random effect model showed a significant medium negative correlation between social support and TI among nurses (p < 0.001). The sample size and TI measurement tools significantly moderated the correlation between social support and TI (p < 0.050). However, nurse department, gender, data collection time, and social support measurement tools did not moderate the correlation between the two variables. Social support is negatively associated with TI in nurses. Nursing administrators and the medical community should fully recognize the importance of social support for nurses and take corresponding measures to enhance it, thereby reducing TI and ensuring the stability of the nursing team.

When are Populist Parties Populist? A Study of Populist Sentiment in Election Manifestos of Germany’s AfD and Die Linke

ABSTRACT In recent years, interest has grown in using election manifestos to measure populism. This, however, is not without its controversies, with critics arguing that party manifestos is an unlikely case for observing populist language. This paper discusses conceptual challenges associated with measuring populism in text-based data, specifically manifestos. Based on previous research, I argue that not all parts of manifestos are equally likely to contain populist language and that variation might be based on the issue being discussed. To test this point, I present a novel dataset containing 32 hand-coded election manifestos of German populist parties AfD and Die Linke. By utilising this data, I analyse the variation in populist sentiment within these manifestos and demonstrate that issues vary in their degree of populism employed and that this is partially related to what issues are salient to populist parties.

Semantic network analysis in consumer and marketing research: application areas in phygital contexts

PurposeLarge-scale text-based data increasingly poses methodological challenges due to its size, scope and nature, requiring sophisticated methods for managing, visualizing, analyzing and interpreting such data. This paper aims to propose semantic network analysis (SemNA) as one possible solution to these challenges, showcasing its potential for consumer and marketing researchers through three application areas in phygital contexts.Design/methodology/approachThis paper outlines three general application areas for SemNA in phygital contexts and presents specific use cases, data collection methodologies, analyses, findings and discussions for each application area.FindingsThe paper uncovers three application areas and use cases where SemNA holds promise for providing valuable insights and driving further adoption of the method: (1) Investigating phygital experiences and consumption phenomena; (2) Exploring phygital consumer and market discourse, trends and practices; and (3) Capturing phygital social constructs.Research limitations/implicationsThe limitations section highlights the specific challenges of the qualitative, interpretivist approach to SemNA, along with general methodological constraints.Practical implicationsPractical implications highlight SemNA as a pragmatic tool for managers to analyze and visualize company-/brand-related data, supporting strategic decision-making in physical, digital and phygital spaces.Originality/valueThis paper contributes to the expanding body of computational, tool-based methods by providing an overview of application areas for the qualitative, interpretivist approach to SemNA in consumer and marketing research. It emphasizes the diversity of research contexts and data, where the boundaries between physical and digital spaces have become increasingly intertwined with physical and digital elements closely integrated – a phenomenon known as phygital.

Particle verbs versus simplex verbs in Maldivian English

AbstractNotwithstanding current indications that English might soon become the country's dominant L1, the structures of English in the Maldives have—despite laudable exceptions—not yet received sufficient academic attention. The present paper studies the contrast between simplex verbs (e.g. COMPRISE) and their innovative particle–verb alternatives (e.g. COMPRISE of) (a) in a short‐term diachronic database of Maldivian newspaper texts and (b) with the help of acceptability ratings of the forms concerned collected through an online survey. The research questions address the relative frequencies of use of innovative particle verbs as opposed to corresponding simplex verbs, the diachronic variability of these relative frequencies and the degree to which the local speech community accepts innovative verb–particle combinations. While the well‐established simplex verbs constitute the overall dominant structural choices, each of the innovative particle verbs has become more frequent in the course of the 14‐year time period observed. These corpus‐based insights align with the generally high acceptability ratings for the innovative particle–verb constructions evident from the survey data, where young women display the largest degree of acceptance towards the innovative particle verbs.

Human Resource Management and Generative Artificial Intelligence (ChatGPT): Nexus, Perspectives and Praxis

The workplace is changing, ChatGPT and other AI technologies are leading the way. ChatGPT is a cutting-edge chatbot created by OpenAI that uses natural language processing to respond to user requests with "human-like" speech. Businesses all over the world are taking notice of it because of its robust capabilities, which have the possibility of helping automate a multitude of company procedures. The aim of the study is to identify the ChatGPT - Nexus, Perspectives and Praxis and investigate the concept’s applicability to generative AI and HRM. This is to underscore and device remediation strategies with respect to likely difficulties in the broad use of ChatGPT in HRM. To stop unfair behaviors and ensure that employees are treated fairly, ethical issues like algorithmic bias and data privacy must be thoroughly examined. The methodology employed involves the use of qualitative paradigm involving the review of literature such as books, journals, publications, articles, online resources among others. The integrative literature review approach enables the researcher to conduct a critical assessment of AI technology vis-à-vis HRM. The study is anchored by one of the most widely accepted theories in human resource management – The Resource-Based View. The study predicts a significant decline in the likelihood of attaining a long-term competitive advantage through strategic Human Resource Management when considering the effects of widely used ChatGPT. The study's findings indicate that ChatGPT help team members communicate and work together more effectively by simplifying the understanding and tracking of project progress, processing and analyzing large amounts of text-based data, facilitating cross-cultural communication, and giving prompt, accurate answers to frequently asked questions. Hiring and recruiting are two critical HR processes towhich generative AI impacts immediate benefits.

A Balancing Act: Data Protection Compliance of Artificial Intelligence

Abstract Neural network based generative artificial intelligence (GenAI) systems, in particular large language models (LLMs), hold huge potential for economic, scientific, and artistic progress. However, they are also the subject of ongoing fierce data protection debates. Some even doubt whether GenAI systems can, at least in principle, comply with the General Data Protection Regulation (EU) 2016/679 (GDPR). This issue was highlighted by a temporary ban of ChatGPT by the Italian data protection authority in March 2023. To answer the question of possible compliance with the GDPR, this paper will first illustrate relevant technical aspects and then examine the extent to which GenAI systems process personal data. A distinction will be made between training, use, and the system itself as stored on a hard drive. On this basis, this paper addresses the key question of the lawfulness of the respective processing activities, bearing in mind that each specific use case of any GenAI system requires a specific data protection assessment. While the issues discussed in this paper are relevant to GenAI systems in general, they are particularly relevant to LLMs trained on large amounts of text-based personal data. As one result, the analysis identifies serious challenges in the application of the GDPR to LLMs. However, with no significant updates to the relevant GDPR provisions in sight, this paper aims to show how pragmatic solutions are both necessary and largely possible even under the current legal framework. This includes practical guidance for developers and users on how to ensure compliance.

Processing 2D barcode data with metaheuristic based CNN models and detection of malicious PDF files

Portable Document Format (PDF) is a file format created to create portable and printable documents across platforms. PDF files are one of the most widely used application types in computer-based systems. Thanks to the functionality that PDF files provide, they are used by many users around the world. Malware developers can exploit PDF files due to various factors. Malware can integrate embedded files, JavaScript, PDF files, etc. As a result, PDFs are susceptible to security vulnerabilities in computer-based systems. In this study, we utilised the CIC-Evasive-PDFMal2022 dataset, made accessible by the Canadian Cybersecurity Institute in 2022, that includes two categories, namely benign and malicious. In the preprocessing step, the proposed model transformed text-based PDF parameter data into the 2D PDF417 barcode. 2D Convolutional Neural Network (CNN) models (MobileNetV2, ResNet18, and ShuffleNet) are trained using the dataset generated by the preprocessing step. CNN is a type of artificial neural network used in image recognition, processing, and classification. Type/class based feature sets were then obtained by each CNN model. In the last step, the metaheuristic optimization method (Honey Badger Algorithm) was used. Thanks to this method, the best performing feature set was determined among the feature sets of the types extracted from each CNN model. It was then classified by the softmax method, and an overall accuracy of 99.73% was achieved. The proposed approach has successfully trained 1D data with 2D CNNs. In addition, with the barcode imaging technique, direct understanding of the data by the users is prevented.

Does the Risk of Exposure to Marijuana Outweigh the Benefits of Breastfeeding? A Systematic Review

Little is known about the risks associated with marijuana use by mothers during the breastfeeding period. A search of CINAHL, Cochrane, Embase, ProQuest Nursing, and PubMed was conducted using a variety of text words and database MeSH terms, including cannabis, medical marijuana, lactation, breastfeeding, marijuana abuse, and marijuana smoking. Articles included both human and animal studies, and there were no language or date restrictions. Hand search and reference lists were also used to access grey literature or any other important articles. Six articles met the inclusion criteria (n = 3 human studies and n = 3 animal studies). In four of the six studies meeting our inclusion criteria, maternal marijuana exposure was associated with detrimental effects, most notably decreased motor development of children at one year of age and lasting neurodevelopmental changes in rats. Breastfeeding mothers should be advised to reduce or cease their marijuana use completely. This article has been peer reviewed.

Exploring corpus linguistics via computational tool analysis: key finding review

Corpus linguistics investigates language using extensive text databases. Tools assist researchers in analyzing, extracting, and interpreting linguistic information efficiently. Furthermore, if researchers only use traditional tools in corpus linguistic analysis, they will lack the comprehensiveness and efficiency required to effectively navigate and derive valuable insights from language data. This paper employed the preferred reporting items for systematic reviews and meta-analyses (PRISMA) approach to find the primary data based on a few keywords in corpus linguistic, corpus analysis, computational linguistic, text corpora and tool support. Based on this method, we used advanced searching techniques on Scopus and Web of Science (WoS) and discovered (N=28) data pertinent to the study. Expert scholars decide on a theme based on the problem, which is (i) types of corpus tools and their uses; (ii) their contributions and their capabilities (iii) limitations of corpus tools. All the tools were used in interdisciplinary studies. In summary, this systematic review uncovers pivotal key findings at the intersection of computational tools and corpus analysis, enriching linguistic knowledge. It highlights the interdisciplinary potential of corpus-based analysis in advancing linguistic tools and, their applications, as well as language analysis.

Effects of Telemedicine on Informal Caregivers of Patients in Palliative Care: Systematic Review and Meta-Analysis.

Telemedicine technology is a rapidly developing field that shows immense potential for improving medical services. In palliative care, informal caregivers assume the primary responsibility in patient care and often face challenges such as increased physical and mental stress and declining health. In such cases, telemedicine interventions can provide support and improve their health outcomes. However, research findings regarding the use of telemedicine among informal caregivers are controversial, and the efficacy of telemedicine remains unclear. This study aimed to evaluate the impacts of telemedicine on the burden, anxiety, depression, and quality of life of informal caregivers of patients in palliative care. A systematic literature search was conducted using the PubMed, Embase, Web of Science, CENTRAL, PsycINFO, CINAHL Plus with Full Text, CBM, CNKI, WanFang, and VIP databases to identify relevant randomized controlled trials published from inception to March 2023. Two authors independently screened the studies and extracted the relevant information. The methodological quality of the included studies was assessed using the Cochrane risk-of-bias tool. Intervention effects were estimated and sensitivity analysis was conducted using Review Manager 5.4, whereas 95% prediction intervals (PIs) were calculated using R (version 4.3.2) and RStudio. A total of 9 randomized controlled trials were included in this study. The meta-analysis indicated that telemedicine has reduced the caregiving burden (standardized mean differences [SMD] -0.49, 95% CI -0.72 to -0.27; P<.001; 95% PI -0.86 to -0.13) and anxiety (SMD -0.23, 95% CI -0.40 to -0.06; P=.009; 95% PI -0.98 to 0.39) of informal caregivers; however, it did not affect depression (SMD -0.21, 95% CI -0.47 to 0.05; P=.11; 95% PI -0.94 to 0.51) or quality of life (SMD 0.35, 95% CI -0.20 to 0.89; P=.21; 95% PI -2.15 to 2.85). Although telemedicine can alleviate the caregiving burden and anxiety of informal caregivers, it does not significantly reduce depression or improve their quality of life. Further high-quality, large-sample studies are needed to validate the effects of telemedicine. Furthermore, personalized intervention programs based on theoretical foundations are required to support caregivers.

A Comprehensive Analysis of Exploring the Efficacy of Machine Learning Algorithms in Text, Image, and Speech Analysis

Our study demonstrates a comprehensive investigation of multiple machine learning models based on text, image, and speech analysis. More specifically, concerning text analysis, we studied the application of such models as Recurrent Neural Networks, Gated Recurrent Units, and Bidirectional Encoder Representations from Transformers in classifying documents into pre-defined categories. The outcomes revealed that RNNs observed the highest precision, recall, F1 score, as well as the highest accuracy. Importantly, the models are well-suited to detect sequential dependencies and create semantic representations on the basis of textual data. With regard to image analysis, we found out that Convolutional Neural Networks were the best model. At the same time, VGG16 and GANs also demonstrated rather promising results suggesting that deep learning is paramount to extract significant data features. As far as speech analysis is concerned, we found out that CNNs are exceptional in terms of accuracy to recognize speech patterns in comparison with the other models. Simultaneously, LSTM also observed a high level of accuracy allowing to capture temporal dependencies in audio signals. In conclusion, the findings of our study suggest that it is exigent to identify an appropriate machine learning model depending on the task and the selected dataset. It is also crucial to understand the nature of each of the studied models to assess their applicability for a specific task. Moreover, our study might be valuable for other researchers given that it contributes to the development of the field of deep learning and, thus, promotes the emergence of new applications in different domains.

Words, Sense, and Denotation Meaning in Pupuh Ginanti “Saking Tuhu Manah Guru”

This study aims to analyze words, sense, and denotation in Pupuh Ginanti, entitled "Saking Tuhu Manah Guru". Pupuh is one of the Balinese poem that is still preserved nowadays. This pupuh has a moral value that needs to be conveyed to listeners. The message is conveyed through the form of the words that compose it. Every word or sentence in a literary work has meant that unknowingly, so an analysis of the word, sense, and meaning of denotation in Pupuh Ginanti needs to be discussed more deeply. Data was initially categorised for this study using some kind of word classification. Start data coding after categorization. By analysing text-based data, data analysis techniques employ qualitative methods. The researcher also employed documentation strategies to obtain the data. The theory used in this study is (Allan, 2014) to analyze sense meaning and supported by theory from (Kreidler, 2002) to analyze denotation meaning. This study uses a descriptive qualitative approach. The result of this research is that the meaning of the poem in Pupuh Ginanti entitled "Saking tuhu manah guru" has a denotative meaning and sense meaning in each line. Sense meaning in this poem was used to clarify the message to be conveyed by the author.

Corporate news disclosure on intangible assets during the COVID-19 crisis in pharmaceutical companies

Purpose: This study explores corporate news communication about intangible assets during the COVID-19 crisis in the largest global pharmaceutical companies. Methodology/approach: Computerised lexical content analysis was performed on 297 articles (texts) from 34 companies, amounting to 280,246 words. We generated a bottom-up clustering of the keywords, titles and abstracts. The qualitative data were obtained from the ProQuest textual database by Clarivate. Findings: The research demonstrates that the biggest global pharmaceutical companies focused intensively on intangible assets during the crisis. They disclosed information on three main intangibles: (1) brand, (2) patent and (3) license, covering 78.17% of the entire corpus text. Research limitations: The study limitations include the fact that the sample concerns only the biggest pharmaceutical companies that met our criteria. Further research could cover small and medium-sized companies and other industries. Finally, further analysis could combine quantitative and qualitative methods within the same research. Originality/value: The research article contributes to the current literature on intangible asset narratives, showing how the biggest pharmaceutical companies tend to achieve a competitive advantage, stay successful during a crisis and address messages to their key stakeholders. Considering the high risks and existing uncertainties, these messages could be of the utmost importance.

Diverse Person: Customize Your Own Dataset for Text-Based Person Search

Text-based person search is a challenging task aimed at locating specific target pedestrians through text descriptions. Recent advancements have been made in this field, but there remains a deficiency in datasets tailored for text-based person search. The creation of new, real-world datasets is hindered by concerns such as the risk of pedestrian privacy leakage and the substantial costs of annotation. In this paper, we introduce a framework, named Diverse Person (DP), to achieve efficient and high-quality text-based person search data generation without involving privacy concerns. Specifically, we propose to leverage available images of clothing and accessories as reference attribute images to edit the original dataset images through diffusion models. Additionally, we employ a Large Language Model (LLM) to produce annotations that are both high in quality and stylistically consistent with those found in real-world datasets. Extensive experimental results demonstrate that the baseline models trained with our DP can achieve new state-of-the-art results on three public datasets, with performance improvements up to 4.82%, 2.15%, and 2.28% on CUHK-PEDES, ICFG-PEDES, and RSTPReid in terms of Rank-1 accuracy, respectively.

Privacy-preserving data integration and sharing in multi-party IoT environments: An entity embedding perspective

The increasing prevalence of IoT applications highlights the urgency for insightful data fusion and information acquisition, boosting data integration and sharing needs. However, challenges arise in multi-party data sharing due to inherent data heterogeneity and privacy concerns. To address these issues, this paper discusses the feasibility of using embedding vectors as the semantic representation, aiming to enhance interoperability across diverse data sources and lay the foundation for natural language-based data querying. At the specific method level, this paper proposes an improved entity tree embedding algorithm to reduce information loss and ameliorate the representation of entity semantics. Additionally, a privacy preservation mechanism based on the entity embedding approach is introduced to provide privacy protection for text-based data. Experimental results on address data demonstrate the mechanism’s efficacy in achieving privacy protection comparable to the widely adopted 2D Laplace plane noise method. Furthermore, incorporating the entity tree embedding into the privacy mechanism could yield more robust and reasonable results regarding location privacy and service quality, signifying the validity of the entity embedding results.

Deciphering Public Voices in the Digital Era

Problem, research strategy, and findings Planners are increasingly using online public engagement approaches to broaden their reach in communities. This results in substantial volumes of digital, text-based public feedback data, making it difficult to analyze efficiently and derive meaningful insights. We explored the use of the novel large language model (LLM), ChatGPT, in analyzing a public feedback data set collected via online submissions in Hamilton City (New Zealand) in response to a proposed local plan change. Specifically, we initially employed zero-shot prompts with ChatGPT for tasks like summarizing, topic identification, and sentiment analysis and compared the results with those obtained by human planners and two standard natural language processing (NLP) techniques: latent Dirichlet allocation (LDA) topic modeling and lexicon-based sentiment analysis. The findings show that zero-shot prompting effectively identified political stances (accuracy: 81.7%), reasons (87.3%), decisions sought (85.8%), and associated sentiments (94.1%). Although subject to several limitations, ChatGPT demonstrates promise in automating the analysis of public feedback, offering substantial time and cost savings. In addition, few-shot prompting enhanced performance in more complex tasks, such as topic identification involving planning jargon. We also provide insights for urban planners to better harness the power of ChatGPT to analyze citizen feedback. Takeaway for practice ChatGPT presents a transformative opportunity for planners, particularly those dealing with growing volumes of public feedback data. However, it cannot be entirely relied upon. Planners must be mindful of ChatGPT’s limitations, including its sensitivity to prompt phrasing, inherent biases from training data, tendency to overgeneralize, and occasional omission of nuanced details. To enhance accuracy, planners should prescreen data for consistency, provide clear and iteratively tested prompts, use few-shot prompts for complex analysis, and explore various combinations of prompting strategies to develop an effective local approach. It is also crucial to ensure human review of the results.

Temporal and spatial distribution characteristics of the Ming Great Wall

The Great Wall during the Ming Dynasty (1368–1644) was the most prominent engineering volume and the most perfect military defense structure in the history of China. By combining traditional historical textual research, site surveys, and GIS database analysis, this paper reveals the Ming Great Wall’s spatial and temporal distribution characteristics and the immediate causes of their formation are discussed in the nine military districts of the frontier. The results show that: (1) The general rule of the Ming Great Wall was to set up military forts, passes, piers, and other related defensive facilities before constructing the continuous wall. (2) The Great Wall spatial and temporal distribution of the phases is relatively apparent, divided into five construction phases in the Ming Dynasty (1368–1620): Hongwu to Xuande (1368–1435), unbuilt the wall; Zhengtong to Tianshun (1436–1464), the consecutive wall appeared; Chenghua to Zhengde (1465–1521), the first construction climax of the Great Wall; Jiajing period (1522–1566), the second construction climax of the Great Wall; Longqing to Wanli (1567–1620), the completion of the whole line of the Great Wall.(3)The warfare and defense posture was the immediate factor that prompted the construction of the Great Wall between the Ming Dynasty and Mongolia. This study has some reference significance for restoring the integrity and authenticity of the Ming Great Wall, and contribute to the preservation of the cultural heritage of.

What methods are used to examine representation of mental ill-health on social media? A systematic review

There has been an increasing number of papers which explore the representation of mental health on social media using various social media platforms and methodologies. It is timely to review methodologies employed in this growing body of research in order to understand their strengths and weaknesses. This systematic literature review provides a comprehensive overview and evaluation of the methods used to investigate the representation of mental ill-health on social media, shedding light on the current state of this field. Seven databases were searched with keywords related to social media, mental health, and aspects of representation (e.g., trivialisation or stigma). Of the 36 studies which met inclusion criteria, the most frequently selected social media platforms for data collection were Twitter (n = 22, 61.1%), Sina Weibo (n = 5, 13.9%) and YouTube (n = 4, 11.1%). The vast majority of studies analysed social media data using manual content analysis (n = 24, 66.7%), with limited studies employing more contemporary data analysis techniques, such as machine learning (n = 5, 13.9%). Few studies analysed visual data (n = 7, 19.4%). To enable a more complete understanding of mental ill-health representation on social media, further research is needed focussing on popular and influential image and video-based platforms, moving beyond text-based data like Twitter. Future research in this field should also employ a combination of both manual and computer-assisted approaches for analysis.

Text Analytics: "Graphic Visualization in Education and a Scientometric Analysis Using R Tool to Explore the Impact and Trends in Classroom Learning."

Text analytics can be applied in education to gain insights into student learning, teacher performance, and educational trends by analysing various forms of text-based data such as student essays, social media posts, and educational resources. While the need for teacher preparation and training is identified, specific solutions or effective strategies for equipping educators with the skills and knowledge to integrate graphic visualizations into their teaching practices might be underexplored. This Scientometrics study used PRISMA approach (Preferred Reporting Items for Systematic Review and Meta-Analysis) Using R tool. Scopus databases that were relevant to the research topic, and its standardized procedures ensured comprehensiveness and replicability. The study found 85 articles using the PRISMA approach from the Scopus database, with 98.82% being articles and 1.18% from conference proceedings. The number of articles published increased in 2022, with 33 articles, compared to 22 articles in 2020. The total number of cited papers was 70, with a total citation of 969. This indicates an increase in research related to graphic visualization in education due to technological advancements. The citation per cited paper was 119. Besides, the study identifies the top 10 countries that have conducted research in graphic visualization, with the United States leading with 19 studies followed by China with 12 and Spain with 8. The top three subject areas in the current research trends are Computer Science education with 78 publications, Engineering education with 33, and Social Science education with 14. Other fields related to language and linguistics are also being given special attention. These findings suggest that education systems and research in education are keeping up with global developments. There is potential for visualizations to be integrated with AI technologies to provide personalized learning experiences, while ensuring that visualizations are accessible to all students, including those with visual impairments. Studies have shown that using these approaches in education can enhance student learning and understanding of complex concepts.

Does heterogeneous media sentiment matter the ‘green premium’? An empirical evidence from the Chinese bond market

This paper selects 346 green bonds issued in China from 2016 to 2021 as the sample, and the Propensity Score Matching (PSM) method is employed to confirm the existence of ‘green premium’ in the Chinese bond market. On this basis, data on internet media sentiment and print media sentiment are collected from ‘Sina Weibo’ and ‘China Important Newspaper Full Text Database’ by both Web Crawler Technology and Textual Analysis Methods to explore the impact and the mechanism of heterogeneous media sentiments on the ‘green premium’. The results show that both the optimism of internet media and print media can significantly promote the ‘green premium’ of green bonds, and the influence of print media sentiment on the ‘green premium’ is greater than that of internet media sentiment. In addition, the Bootstrap method verifies the mediating effect of print media sentiment in the influence of internet media sentiment on ‘green premium’, indicating that print media sentiment is an important transmission path. Moreover, the results of the heterogeneity test show that the more optimistic the media is, the more significant the ‘green premium’ effect is in the regions with higher institutional environments and financial subsidies. The ‘green premium’ of green bonds is most pronounced for higher levels of institutional environment and green bond preferential policies.