News Corpus Research Articles

Fake news detection and corpus establishment from comment data for social network posts

Fake news detection and corpus establishment from comment data for social network posts

Uzbek news corpus for named entity recognition

Uzbek news corpus for named entity recognition

The news values of fake news

Fake news poses a significant threat to society by undermining public trust and consensus on critical issues. Although there is a considerable amount of research on the linguistic features of fake news texts, a comprehensive understanding of how language is used to persuade and promote specific ideologies within them is still lacking. This study addresses this gap by analyzing fake news discourse through the lens of news values. We apply the Discursive News Values Analysis (DNVA) framework and key semantic domain analysis to a corpus of fake news stories on vaccination, climate change, and COVID-19. We identify a set of news values that differentiate fake from mainstream news discourse. Our findings reveal that fake news emphasizes negativity, unexpectedness, consonance, and facticity, while also relying on the previously undocumented news values of subversiveness, causality, religiosity, and historicity. These values form a powerful discursive toolkit exploited by fake news writers to craft compelling false narratives.

Coverage of Political Unrest in Pakistani English Newspapers: A Corpus-Based Content Analysis

The newspaper is a traditional channel of information interchange between the public and the government, as well as a free platform for the general public to commend and criticize the actions and policies of the government. This article aims to study the frequency of lexical choices and positive and negative words in the news coverage of political unrest by Pakistani English newspapers using corpus-based content analysis. The data for analysis was collected from the official websites of three Pakistani English newspapers: Daily Times, The Nation, and Dawn. The collected data has been analyzed quantitatively using the corpus software AntConc. Twenty news items were selected for news corpus with purposive sampling from 5th April 2023 to 30th September 2023. The findings reveal that Pakistani English newspapers use more negative words in the coverage of political matters, which do not represent the subjectivity of the newspapers' publishers but provide trustworthy news about political uncertainty to the audience. Moreover, this study explores the agenda of Pakistani English newspaper media through the Agenda Setting theory by Maxwell McCombs and Donald Shaw (1972). Additionally, this study demonstrates how much realistically Pakistani English publications depict political events that are causing uncertainty.

A dual-ways feature fusion mechanism enhancing active learning based on TextCNN

Active Learning (AL) is a technique being widely employed to minimize the time and labor costs in the task of annotating data. By querying and extracting the specific instances to train the model, the relevant task’s performance is improved maximally within limited iterations. However, rare work was conducted to fully fuse features from different hierarchies to enhance the effectiveness of active learning. Inspired by the thought of information compensation in many famous deep learning models (such as ResNet, etc.), this work proposes a novel TextCNN-based Two ways Active Learning model (TCTWAL) to extract task-relevant texts. TextCNN takes the advantage of little hyper-parameter tuning and static vectors and achieves excellent results on various natural language processing (NLP) tasks, which are also beneficial to human-computer interaction (HCI) and the AL relevant tasks. In the process of the proposed AL model, the candidate texts are measured from both global and local features by the proposed AL framework TCTWAL depending on the modified TextCNN. Besides, the query strategy is strongly enhanced by maximum normalized log-probability (MNLP), which is sensitive to detecting the longer sentences. Additionally, the selected instances are characterized by general global information and abundant local features simultaneously. To validate the effectiveness of the proposed model, extensive experiments are conducted on three widely used text corpus, and the results are compared with with eight manual designed instance query strategies. The results show that our method outperforms the planned baselines in terms of accuracy, macro precision, macro recall, and macro F1 score. Especially, to the classification results on AG’s News corpus, the improvements of the four indicators after 39 iterations are 40.50%, 45.25%, 48.91%, and 45.25%, respectively.

Amina: an Arabic multi-purpose integral news articles dataset

Electronic newspapers are one of the most common sources of Modern Standard Arabic. Existing datasets of Arabic news articles typically provide a title, body, and single label. Ignoring important features, like the article author, image, tags, and publication date, can degrade the efficacy of classification models. In this paper, we propose the Arabic multi-purpose integral news articles (AMINA) dataset. AMINA is a large-scale Arabic news corpus with over 1,850,000 articles collected from 9 Arabic newspapers from different countries. It includes all the article features: title, tags, publication date and time, location, author, article image and its caption, and the number of visits. To test the efficacy of the proposed dataset, three tasks were developed and validated: article textual content (classification and generation) and article image classification. For content classification, we experimented the performance of several state-of-the-art Arabic NLP models including AraBERT and CAMeL-BERT, etc. For content generation, the reformer architecture is adopted as a character text generation model. For image classification applied on Al-Sharq and Youm7 news portals, we have compared the performance of 10 pre-trained models including ConvNeXt, MaxViT, ResNet18, etc. The overall study verifies the significance and contribution of our newly introduced Arabic articles dataset. The AMINA dataset has been released at https://huggingface.co/datasets/MohamedZayton/AMINA.

Real-Time Extraction of News Events Based on BERT Model

Abstract For the large number of news reports generated every day, in order to obtain effective information from these unstructured news text data more efficiently. In this paper, we study the real-time crawling of news data from news websites through crawling techniques and propose a BERT model-based approach to extract events from news long text. In this study, NetEase news website is selected as an example for real-time extraction to crawl the news data of this website. BERT model as a pre-trained model based on two-way encoded representation of transformer performs well on natural language understanding and natural language generation tasks. In this study, we will fine-tune the training based on BERT model on news corpus related dataset and perform sequence annotation through CRF layer to finally complete the event extraction task. In this paper, the DUEE dataset is chosen to train the model, and the experiments show that the overall performance of the BERT model is better than other network models. Finally, the model of this paper is further optimised, using the ALBERT and RoBERTa models improved on the basis of the BERT model, experiments were conducted, the results show that both models are improved compared to the BERT model, the ALBERT model algorithm performs the best, the model algorithm's F1 value is 1% higher than that of BERT. The results show that the performance is optimised.

Public Behavior and Emotion Correlation Mining Driven by Aspect From News Corpus.

Emotion motivates behavior. Investigating the correlation between behavior and emotion, an often overlooked perspective, plays a significant role in uncovering the underlying motives behind behaviors and the intrinsic cause-effects of social events. This article proposes a methodology for mining the correlation between public behavior and emotion using daily news data. Initially, aspect-emotion-reaction (A-E-R) triplets are extracted and generalized, encompassing both explicit and implicit patterns. Then, a knowledge representation model based on hypothetical context (KRHC) with a self-reflection mechanism is proposed to uncover implicit relationships between emotion and behavior through attention mechanisms. By combining rule-based methods for explicit relationships and deep learning for implicit ones, an understanding of emotion-behavior patterns is achieved. In this study, the behaviors are divided into three categories of prosocial, antisocial, and normal behaviors with ten secondary types. Seven categories of emotions are adopted. The proposed deep learning model KRHC is validated on A-E-R datasets and public KINSHIP datasets. The experiment results are concluded; for example, when "fear", "sad", and "surprise" emotions appear, it drives behavior "panic" with most probability. These findings could provide insights for both human-computer interaction and public safety management applications.

Forecasting Inflation Using Economic Narratives

We use economic narratives to forecast inflation with a large news corpus and machine learning algorithms. The economic narratives from the full text content of over 880,000 Wall Street Journal articles are decomposed into multiple time series representing interpretable news topics, which are then used to predict inflation. The results indicate that narrative-based forecasts are more accurate than the benchmarks, especially during recession periods. Narrative-based forecasts perform better in long-run forecasting and provide incremental predictive information even after controlling macroeconomic big data. In particular, information about inflation expectations and prices of specific goods embedded in narratives contributes to their predictive power. Overall, we provide a novel representation of economic narratives and document the important role of economic narratives in inflation forecasting.

Indonesian disaster named entity recognition from multi source information using bidirectional LSTM (BiLSTM)

Precise logistic support is essential after a disaster occurs. It must be timely, accurate, targeted, and based on existing needs. However, obtaining sufficient and accurate information related to logistic distribution locations remains a key problem. Therefore, implementing Named Entity Recognition (NER) can address this issue. In recent years, news coverage through Indonesian digital news media and social media accounts has emerged as a promising source for building a disaster data corpus. This study implemented NER to extract and identify named entities from text-based information, particularly from Indonesian digital news media. In addition to using regular entities from the NER standard, this study introduced new entities specialized for disaster-related information, including DISASTER, SCALE, SUPPLIES, CASUALTIES, and OUTSIDE. The new disaster corpus in the Indonesian language for the NER model was obtained with an imbalanced dataset composition. To overcome this problem, random oversampling was applied. This study also utilized the BiLSTM model to recognize each entity in new textual information, evaluating its performance when the proposed Indonesian disaster corpus was used as a training reference in the deep learning model. Several optimization algorithms applied in BiLSTM were evaluated. The results showed improved BiLSTM performance using Adam optimization and a balanced corpus. Performance indicators achieved were 93.4 %, 82.4 %, and 87.5 % for precision, recall, and F1-score, respectively. The BiLSTM network captured long-range dependencies in sequential data provided by NER. Oversampling ensured that the proposed NER model could precisely recognize all entities and reduce biased results. Thus, the BiLSTM method can better identify entities in the textual corpus of Indonesian disaster-related online news.

Shame-Sensitive Public Health.

In this article, we argue that shaming interventions and messages during Covid-19 have drawn the relationship between public health and shame into a heightened state of contention, offering us a valuable opportunity to reconsider shame as a desired outcome of public health work, and to push back against the logics of individual responsibility and blame for illness and disease on which it sits. We begin by defining shame and demonstrating how it is conceptually and practically distinct from stigma. We then set out evidence on the consequences of shame for social and relational health outcomes and assess the past and present dimensions of shame in the context of the Covid-19 pandemic, primarily through a corpus of international news stories on the shaming of people perceived to have transgressed public health directions or advice. Following a brief note on shame (and policymaking) in a cultural context, we turn to the concept and practice of 'shame-sensitivity' in order to theorise a set of practical and adaptable principles that could be used to assist policymakers in short- and medium-term decision-making on urgent, tenacious, and emerging issues within public health. Finally, we consider the longer consequences of pandemic shame, making a wider case for the acknowledgement of the emotion as a key determinant of health.

Corpus-based Multi-dimensional Analysis of the Register Features of Financial News

Nowadays, people rely heavily on financial news media to get financial updates. But as is well known, for non-native speakers of English, in order to fully understand such news, they also need to learn some register features of such financial news besides adequate English language knowledge. The purpose of this study is to explore and analyze the linguistic features of English financial news, and to help readers effectively understand financial news. Specifically, this study employes multi-feature/multi-dimensional analysis (henceforth MF/MD) method to analyze the register characteristics of English financial news based on the theory of register. Through analyzing the self-compiled corpus of financial news, it is found that that the financial news has four major register features: interactivity, persuasion, information and professionalism. These findings will shed lights on the understanding of financial news as well as the learning and teaching of financial/business English.

The Conceptual Metaphor PARENTS ARE ANIMALS: On Linguistic Terms Used Figuratively for Types of Parenting

The aim of this paper is to provide an analysis of English animal-related terms used figuratively with reference to types of parenting. To be more precise, we make an attempt to investigate the scope of the conceptual metaphor PARENTS ARE ANIMALS. What ignited our interest was the ubiquity of linguistic metaphors featuring animal terms employed with reference to types of parenting present in everyday language, thus we have decided to delve into what types of animals are typical source domains in metaphors of this type. The methodology adopted is the Conceptual Metaphor Theory, originally proposed by Lakoff and Johnson (1980) and further developed in numerous subsequent publications, especially those by Kövecses (2003, 2005, 2010, 2015, 2017a, 2017b). Our results are supported by evidence from the online Corpus of Contemporary American English and News on the Web Corpus, which may be said to corroborate not only the conceptual nature of metaphors as such, but also their impact on social cognition.

Aesthetic Plastic Surgery Issues During the COVID-19 Period Using Topic Modeling

This study investigates media coverage of cosmetic surgery in South Korea from 2014 to 2023 using text mining techniques applied to news articles from BigKinds. It focuses on assessing the prevalence of objective information and the societal impacts of capital-driven misinformation. The research methodology involved optimal topic modeling through perplexity, likelihood, BIC, and similarity measures, identifying five themes within the cosmetic surgery news corpus. Further analysis included quantitative topic recognition via fuzzy clustering by period, sentiment analysis, and network analysis utilizing n-gram techniques to explore relationships between key terms. Findings reveal five main topics covered in cosmetic surgery news: Consumer Psychology, Cosmetic Surgery Market, Cosmetic Companies and Technologies, Side Effects and Incidents, and the Tourism Industry. The period from 2014 to 2016 saw significant coverage, particularly on medical tourism and surgical side effects, while in 2017, attention shifted to the surgical process and market stability. From 2018 onward, news coverage expanded, especially in May, focusing on cosmetic technology and related industries amid increased outdoor activities. With the COVID-19 pandemic in 2020, there was a resurgence in coverage of the cosmetic surgery market. In 2023, post-pandemic, there was an uptick in articles related to cosmetic surgery technology industries and support funds. The core words in cosmetic surgery news were spreading around "plastic surgery," "China," and "Botulinum". The study sheds light on the potential influence of capital on media portrayals of cosmetic surgery and the resulting societal consequences of misinformation.

Topic Mining and Evolution of U.S. Mainstream Media Reporting on “Belt and Road Initiative” (2013-2023)

The year 2023 marks the 10th anniversary of the the Belt and Road Initiative. A comprehensive analysis and evaluation of mainstream media reports in the United States offer valuable insights into the overall understanding and strategic trends of the United States over the past decade. Such insights hold significant importance for shaping the future development of the Belt and Road Initiative. This study utilizes Factiva’s extensive self-compiled “Belt and Road Initiative” news corpus, employing social network analysis, multiple correspondence analysis, and co-occurrence network analysis to visualize the topic distribution and evolution in American media reports on Belt and Road Initiative. The identified topics in American “Belt and Road Initiative” reports encompass transportation, finance and trade, diplomacy and politics, science and technology, ecological environment, and medical care and health. From a diachronic perspective, the evolution of US “B&R” reporting unfolds in three distinct stages. The period from 2013 to 2015 is characterized by a focus on macro-policy reports, shifting to economic and technological reports from 2016 to 2019. The subsequent stage, spanning 2020 to 2023, centers on ecological sustainability and debt reporting. Furthermore, this study observes that the content of US “B&R” reports exhibits a pronounced political nature, gradually leaning towards characterizing Belt and Road Initiative as China’s geostrategic layout for developing countries.

Limitations of Large Language Models in Propaganda Detection Task

Propaganda in the digital era is often associated with online news. In this study, we focused on the use of large language models and their detection of propaganda techniques in the electronic press to investigate whether it is a noteworthy replacement for human annotators. We prepared prompts for generative pre-trained transformer models to find spans in news articles where propaganda techniques appear and name them. Our study was divided into three experiments on different datasets—two based on an annotated SemEval2020 Task 11 corpora and one on an unannotated subset of the Polish Online News Corpus, which we claim to be an even bigger challenge as an example of an under-resourced language. Reproduction of the results of the first experiment resulted in a higher recall of 64.53% than the original run, and the highest precision of 81.82% was achieved for gpt-4-1106-preview CoT. None of our attempts outperformed the baseline F1 score. One of the attempts with gpt-4-0125-preview on original SemEval2020 Task 11 achieved an almost 20% F1 score, but it was below the baseline, which oscillated around 50%. Part of our work that was dedicated to Polish articles showed that gpt-4-0125-preview had a 74% accuracy in the binary detection of propaganda techniques and 69% in propaganda technique classification. The results for SemEval2020 show that the outputs of generative models tend to be unpredictable and are hardly reproducible for propaganda detection. For the time being, these are unreliable methods for this task, but we believe they can help to generate more training data.

A study on deep learning for Vietnamese text classification

Text categorization aims to automatically assign given text passages or documents to predetermined categories or subjects. Despite the wide array of techniques employed in classifying English text, there remains a dearth of research on Vietnamese text classification. This paper introduces a novel approach utilizing a Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) with a deep network structure for Vietnamese text classification. Our findings demonstrate a substantial improvement in classification accuracy when applying deep learning techniques to two Vietnamese news corpus datasets. This study contributes to the advancement of Vietnamese text classification by introducing and demonstrating the efficacy of LSTM and CNN with a deeper network structure. The results offer valuable insights for researchers and practitioners working on text categorization in the Vietnamese language.

What if migrants were only people and relatives? Designations used to name people on the move in the Belgian media

This article focuses on denominations that are used to name people on the move in Belgian media discourse, but that are not specifically related to migration. It specifically studies the nominal syntagms formed with the noun people ( people on the run, people in need) and words of kinship ( mother, brother). A Discursive Semantics analysis implemented through Corpus Linguistics is run on a corpus of Belgian news items issued from March 2015 to July 2017. The corpus gathers 13,391 newspaper articles and 3490 TV news items (representing 7,637,986 words). The mention of words of kinship and designations formed with people shows that there is a willingness to humanise media discourses on migration. However, although their mention encourages a humanitarian vision of people on the move, these usually positively connotated designations also foster a vision of people on the move as victims and does not discourage the mention of controversial denominations.

From bonus to burden: The cost of ruling from a new(s) perspective

AbstractStudies have repeatedly documented the cost of ruling: governing parties generally suffer electoral defeats. We approach this empirical law of political science from the perspective of another empirical law: the incumbency bonus, the fact that incumbents get more media attention than the opposition. Our claim is that the bonus constitutes an electoral liability because it reflects the critical approach of media to government power. News featuring incumbents is therefore associated with a more negative tone than news featuring the opposition. This incumbency burden in turn affects government support negatively. Empirically, we draw on an extensive news corpus covering four European countries over two decades, combined with monthly poll data. Analyses show that the incumbency burden in political news is an empirical reality, and that variations in the burden contribute substantially to predictions of government support. Finally, the negative burden effect is stronger for single‐party cabinets, but stable throughout government tenure.

Variasi dan Komponen Makna Verba Pewarta pada Korpus Berita Daring

"Reporting verbs indicate a conversation (Romli, 2018:35). In news texts, reporting verbs are used intensively to provide information. This research focuses on the identification of variations, frequencies, and the analysis of meaning components as proposed by Nida (1975) regarding the usage of reporting verbs in online news corpus. This research uses a qualitative descriptive method with a lexical semantic approach. Data were obtained from news texts published by Kompas.com and Detik.com during August 8-10, 2023, with the theme 'Ferdy Sambo's Death Sentence Turned into Life Imprisonment.' Based on the research findings, 17 variations of reporting verbs were found in the news corpus, with the verb 'mengatakan' (to say) being the most dominant with a frequency of 83 occurrences. Meanwhile, based on the analysis of meaning components for the verbs 'mengatakan'(to say), 'menyatakan' (to state), 'menjelaskan' (to explain), and 'menyebut' (to mention), differences in usage were found through the differentiation of the meaning components of each reporting verb. This research is expected to contribute by highlighting the differences in meaning among synonymous verbs to demonstrate the diversity of the Indonesian language vocabulary.