Text Analyses Research Articles

Changes in menstrual symptoms and work productivity after checklist-based education for premenstrual syndrome: an 8-month follow-up of a single-arm study in Japan.

Premenstrual syndrome (PMS) is prevalent among women of reproductive age, but most do not seek medical advice. We hypothesized that building PMS awareness could promote medical help-seeking for PMS and thus reduce menstrual symptoms and improve work productivity. In January 2020, women aged between 25 and 44years, having paid work, and not currently consulting with an obstetrics and gynecology doctor (n = 3090) responded to the Menstrual Distress Questionnaire (MDQ), the Premenstrual Symptoms screening tool, and the World Health Organisation Health and Work Performance Questionnaire. In addition, they received checklist-based online education for PMS. Of 3090 participants, 2487 (80.5%) participated in a follow-up survey in September 2020. We conducted multiple logistic regression analyses and text analyses to explore factors that encouraged and discouraged medical help-seeking. We also evaluated changes in menstrual symptoms and work productivity, using generalized estimating equations with interactions between the severity of PMS, help-seeking, and time. During the follow-up period, 4.9% of the participants (121/2487) sought medical help. Those having high annual income (adjusted odds ratio [aOR] = 2.07, 95% confidence interval [CI]: 1.21-3.53) and moderate-to-severe PMS (aOR = 2.27, 95% CI: 1.49-3.46) were more likely to have sought medical help. Those who did not seek medical help despite their moderate-to-severe PMS reported normalization of their symptoms (36%), time constraints (33%), and other reasons for not seeking medical help. Participants with moderate-to-severe PMS who had sought medical help showed a significant improvement of - 8.44 points (95% CI: - 14.73 to - 2.15 points) in intermenstrual MDQ scores during the follow-up period. However, there were no significant improvements in premenstrual and menstrual MDQ scores or absolute presenteeism. Medical help-seeking alleviated intermenstrual symptoms in women with moderate-to-severe PMS, but only a small proportion of them sought medical help after PMS education. Further research should be conducted to benefit the majority of women who are reluctant to seek medical help, including the provision of self-care information. UMIN Clinical Trials Registry number: UMIN000038917.

Poczucie własnej skuteczności studentów w rozwijaniu sprawności pisania w języku obcym

Self-efficacy is a well-known and researched concept originating from the social-cognitive theory of Albert Bandura (1977, 1997). However, it still seems to be able to promote interesting reflections that are important for language education in higher education, both in teacher and learner activities. According to a number of studies learners acquire important skills and attitudes that are linked to a high degree of self-efficacy: they are highly motivated to learn, set ambitious goals for themselves, and are more persistent and committed to learning than those with a low degree of self-efficacy. The text analyses the results of a quantitative-qualitative study conducted among students participating in foreign language classes at the University of Warsaw. The collected data is analysed with the aim of answering the following research questions: What is the level of self-efficacy in written expression in a foreign language at B2-C2 levels? How do the students perceive the teacher-lecturer’s actions to strengthen their sense of self-efficacy? Do they see strengthening their involvement in written expression as a key element of self-efficacy? In addition, we are also interested in their opinion towards the possibilities of Chat GPT in this regard. The results are used to propose some teaching activities that will enhance students’ self-efficacy in written expression at B2-C2 levels as an important motivational factor which may result in a sense of agency.

Email Campaigns That Suit the Candidate: Leveraging Automated Text Analysis to Increase Political Donations

This research employs automated text analysis to explore how textual characteristics in campaign emails affect monetary donations received by political candidates. The authors outline a new methodological framework that combines a machine learning approach for natural language processing with fixed effect regressions, thereby enabling researchers to study and interpret the impact of textual characteristics on donations while also accounting for individual differences across candidates and their email recipients. Using this framework, the authors analyze 764 emails from 19 candidates in the 2020 U.S. Democratic presidential primary election and evaluate how certain textual characteristics (e.g., empathy, vulnerability) in campaign emails affect donation outcomes. Identifying these effects would enable candidates to improve their email text and increase their donations by 9% on average. This research provides a practical and flexible roadmap for automated text analysis in situations where political campaigns do not have clear a priori hypotheses about which textual characteristics will be effective for them.

A systematic evaluation of text mining methods for short texts: Mapping individuals' internal states from online posts.

Short texts generated by individuals in online environments can provide social and behavioral scientists with rich insights into these individuals' internal states. Trained manual coders can reliably interpret expressions of such internal states in text. However, manual coding imposes restrictions on the number of texts that can be analyzed, limiting our ability to extract insights from large-scale textual data. We evaluate the performance of several automatic text analysis methods in approximating trained human coders' evaluations across four coding tasks encompassing expressions of motives, norms, emotions, and stances. Our findings suggest that commonly used dictionaries, although performing well in identifying infrequent categories, generate false positives too frequently compared to other methods. We show that large language models trained on manually coded data yield the highest performance across all case studies. However, there are also instances where simpler methods show almost equal performance. Additionally, we evaluate the effectiveness of cutting-edge generative language models like GPT-4 in coding texts for internal states with the help of short instructions (so-called zero-shot classification). While promising, these models fall short of the performance of models trained on manually analyzed data. We discuss the strengths and weaknesses of various models and explore the trade-offs between model complexity and performance in different applications. Our work informs social and behavioral scientists of the challenges associated with text mining of large textual datasets, while providing best-practice recommendations.

Countering the “Climate Cult” – Framing Cascades in Far-Right Digital Networks

ABSTRACT In many contemporary democracies, digital networks on the far-right have established themselves as “alternatives” to liberal institutions. Within this nexus of parties, hyper-partisan news, and social media channels, opposition to climate change mitigation has emerged as a central cross-cutting issue. Despite the prominence of these trends, we still know relatively little about the hierarchies that govern these communicative ties. This study draws on the concept of cascading frame activation to trace communication flows between four distinct actor types within the German far right. We apply a mixed-methods approach across social media and websites over a period of 19 months. First, we conduct a time series analysis via a vector autoregression model, to assess potential causal relationships between the issue agendas of far-right actor types. Automated text analysis then inductively generates a lexicon exclusive to far-right posts about climate change, which we qualitatively assess and manually cluster into distinct frame categories. We conclude by tracing framing cascade sequences across the various actor types. Our results reveal significant associations between the issue agenda of far-right actors, especially among right-wing news sites (RNS) and hyperactive accounts. Within this network, the most dominant frames demarcate out-groups, such as the Green Party or climate activists. Notably, RNS play a central role in shaping this discourse, initiating 42% of the identified frames before they find wider dissemination. These findings expand our understanding of hierarchies and new communication pathways among partisan media, political parties, and their extended digital networks on the far right.

Особенности восприятия научного текста студентами медицинского университета: психолингвистический эксперимент

The article highlights a psycholinguistic experiment performed to describe the process of perception, understanding and generation of a scientific text by students of a medical university. Research texts and scientific communication are media of the institutional discourse and communication. The text, in this case, is addressed in its dynamic state, thus correlating with cognitive processes of understanding and perception. The text, being a powerful tool of speech influence, allows evaluating and analyzing diverse types of cognitive activity of its recipients. The study is of current interest, since, as reported, modern ways of communication disseminated through mass media and social networks result in decreased cognitive and analytical abilities of young people. All of the above determined the purpose and objectives of the study: to describe the process of scientific communication the participants of which are medical university students; to consider features of perceiving a scientific text by this category of readers, to specify models of understanding a scientific text. The experiment consisted of two stages: the stage of material collection using the countertext approach developed by A.I. Novikov, and the stage of processing and analysing the material obtained. To evaluate appropriateness of the stimulus text to the experimental conditions, we analysed the text quantitatively and qualitatively using the RuLingva text analyzer according to the criteria of “readability”, “informativeness”, “abstractness”, “complexity”. The analysis of the material obtained allowed describing the strategies and models used by the medical students in the process of scientific communication. The most frequent reactions were the reactions of a meaningful nature corresponding to the stages of cognitive activity – “paraphrasing”, “statement”, “generalization”, “association”; also new reactions were described: “recommendation”, “disapproval”, “empathy”, “indifference” – their manifestation may be due to the presence of “medical experience” among the participants of the experiment.

Od listopada do listopada. Monarchii habsburskiej i Polaków wzajemne rozstawanie

The author of the text analyses the reasons why Poles from Galicia, southern Poland, departed from the so-called Austro-Polish conception and matured towards the ideal of full independence. He draws attention to the significance of the Act of 5 November 1916 for the initiation of emancipation processes among Poles living under the Habsburg rule.

Learning deep news sentiment representations for macro-finance

AbstractThis paper introduces custom neural network techniques to the problem of latent economic factor extraction for voluminous news analytics data. In the context of macro-financial news, we derive low-dimensional representations of time series that arise in textual sentiment analyses spanning various topics. We explore three applications for compressed news sentiment data: nowcasting GDP growth, explaining asset class returns in a panel data analysis, and time series momentum investment. Our empirical study shows that nonlinear data representations based on supervised autoencoder architectures compare favorably to alternatives across all applications. In specific, we demonstrate that augmenting autoencoders with supervision tasks based on common asset class returns and market characteristics disciplines the dimension reduction and naturally supports the transparency of resulting representations. Taken together, our findings position supervised autoencoders as attractive competitor models alongside PCA and PLS approaches.

What is the place of interpretation in text analysis? An example using ALCESTE® software

ABSTRACT Statistical text analyses appear at the borderline of quantitative and qualitative research, raising epistemological questions in the field. This paper focuses on a specific method of text analysis: the Reinert method. It aims to address the role of interpretation and subjectivity in the construction of the results using this kind of analysis. To this end, this manuscript presents a dual interpretation of a same corpus and software outputs highlighting the convergences and divergences in both researchers’ interpretations. It discusses how the use of statistics on qualitative data may provide a false sense of objectivity and sweep the questions of interpretation and epistemology away. Finally, it offers clear guidelines regarding individual and dual interpretation to those who wish to start using the Reinert method. We argue that this type of analysis is adaptable and can be applied across a variety of epistemological stances and research questions.

The Long March: The quest for valid text-based indicators of exploration and exploitation

Since March outlined the importance of balancing exploration and exploitation in organizational learning, the exploration–exploitation paradigm has received substantial attention in the management literature. Recent studies have used computer-aided text analysis to construct measures of firms’ inclination towards exploration or exploitation, using the original set of keywords proposed by James G. March. We propose a structured series of tests to assess the validity of computer-aided text analysis-based measures and demonstrate that the approach used in prior studies is unlikely to deliver valid indicators. We demonstrate that an alternative approach, which relies on a larger library of keywords, including synonyms of the March keywords and selects only those keywords that are informative and that pass validity tests, delivers valid computer-aided text analysis indicators – both for unstructured (news articles) and structured text-bases (annual reports). Our study contributes to the literature on construct validity and has broader implications for the development of computer-aided text analysis-based indicators in strategy and organization research.

Undergraduate Students’ Definitions of Sexual Violence Labels Across the Rise of the #MeToo Movement

Examining changes in perspectives of sexual violence across the #MeToo movement is critical to understanding shifts in sociocultural contexts of sexual violence; yet the literature is still nascent. In the present study, we assessed variation across the #MeToo movement in undergraduate students’ ( N = 1,028) rape myth acceptance, sexual violence experiences, identification of experiences with sexual violence labels, and self-generated definitions of sexual violence labels (e.g., sexual assault). We also examined the mediating role of the relation between timepoint across #MeToo and students’ self-generated sexual violence definitions. Text analyses revealed seven themes in students’ definitions: Activity, Chronicity, Body, Contact, Gender, Force/Harm, and Consent. Notably, there was greater representation and a higher rate of change in the Contact theme in definitions following the #MeToo movement compared to pre-#MeToo, but only among students with sexual violence history. Rape myth acceptance did not change significantly across the #MeToo movement and did not demonstrate a significant role in time-related variability in students’ definitions. These findings provide support for the role of the #MeToo movement in facilitating more inclusive perceptions of sexual violence among undergraduate students, especially among those with sexual violence histories. We hope that with increased insight into the #MeToo movement's impact on sexual violence perspectives, continued efforts of the movement and prevention initiatives can broaden their reach.

Large-scale text analysis using generative language models: A case study in discovering public value expressions in AI patents

Abstract We put forward a novel approach using a generative language model (GPT-4) to produce labels and rationales for large-scale text analysis. The approach is used to discover public value expressions in patents. Using text (5.4 million sentences) for 154,934 US AI patent documents from the United States Patent and Trademark Office (USPTO), we design a semi-automated, human-supervised framework for identifying and labeling public value expressions in these sentences. A GPT-4 prompt is developed that includes definitions, guidelines, examples, and rationales for text classification. We evaluate the labels and rationales produced by GPT-4 using BLEU scores and topic modeling, finding that they are accurate, diverse, and faithful. GPT-4 achieved an advanced recognition of public value expressions from our framework, which it also uses to discover unseen public value expressions. The GPT-produced labels are used to train BERT-based classifiers and predict sentences on the entire database, achieving high F1 scores for the 3-class (0.85) and 2-class classification (0.91) tasks. We discuss the implications of our approach for conducting large-scale text analyses with complex and abstract concepts. With careful framework design and interactive human oversight, we suggest that generative language models can offer significant assistance in producing labels and rationales.

The Disclosure Tone, Earnings Management, Earnings Quality: Evidence Found in Indonesia Listed Manufacturing Industry

This study examined the influence of earnings management (EM) and earnings quality (EQ) on the disclosure tone of an annual report. NVIVO’s coding AI measures the DT (Disclosure Tone) through automated text analysis. The sample used was 61 of Indonesia’s listed manufacturing companies from 2017 to 2021. The balanced panel data was used and executed by a weighted panel least square. The results prove that EM is essential for DT, while EQ is insignificant for DT. The higher EQ indicates the high asymmetric information from management to the stakeholders, which leads the company to disclose a more positive tone. However, the negative tone is also improving with a higher EQ. These results imply that when the company has higher EM, it tries to balance the tone of its annual report in order to gain the trust of stakeholders, this is beneficial for analysts, investors, and regulators when collecting information from annual reports. Future studies can extend this framework to other types of industries since each industry has specific characteristics in giving the tone in their reports. Besides, future studies can also focus on the tone of the sustainability report or integrated report in other countries.
 Keywords: earnings management, earnings quality, disclosure tone, manufacturing industry

Leveraging distant supervision and deep learning for twitter sentiment and emotion classification

AbstractNowadays, various applications across industries, healthcare, and security have begun adopting automatic sentiment analysis and emotion detection in short texts, such as posts from social media. Twitter stands out as one of the most popular online social media platforms due to its easy, unique, and advanced accessibility using the API. On the other hand, supervised learning is the most widely used paradigm for tasks involving sentiment polarity and fine-grained emotion detection in short and informal texts, such as Twitter posts. However, supervised learning models are data-hungry and heavily reliant on abundant labeled data, which remains a challenge. This study aims to address this challenge by creating a large-scale real-world dataset of 17.5 million tweets. A distant supervision approach relying on emojis available in tweets is applied to label tweets corresponding to Ekman’s six basic emotions. Additionally, we conducted a series of experiments using various conventional machine learning models and deep learning, including transformer-based models, on our dataset to establish baseline results. The experimental results and an extensive ablation analysis on the dataset showed that BiLSTM with FastText and an attention mechanism outperforms other models in both classification tasks, achieving an F1-score of 70.92% for sentiment classification and 54.85% for emotion detection.

Populist Radical-Right Attitudes, Political Involvement and Selective Information Consumption: Who Tunes Out and Who Prefers Attitude-Consonant Information

ABSTRACT This study seeks to understand how populist radical-right (PRR) attitudes and political involvement relate to individuals’ political information consumption and selective exposure to ideological content. The study approaches political information selection as a two-step phenomenon: first, individuals use different amounts of political information, and second, they rely on attitude-consonant information to different degrees. Results from a combination of survey measures, implicit association tests and automated text analysis of large-scale online tracking data collected in Germany and Switzerland in 2020 showed that first, political information consumption was related to political involvement but also to PRR attitudes—in different ways—in both countries. Second, our analyses revealed country variations regarding attitude-consonant online PRR exposure. Third, implicit PRR attitudes were not relevant in explaining the use of PRR content online. We discuss the relevance of moving beyond analyses of traditional news toward the breadth of political information consumption, country contexts, and naturalistic research designs.

‘Emigration is Luck’: Destiny, Witchcraft and Uncertainty in Migratory Journeys from the Gambia and Guinea-Bissau

ABSTRACT This text analyses religious interpretations of the uncertainty of migratory journeys – documented and clandestine – from West Africa to Europe, based on beliefs in Islamic destiny, ‘luck’, witchcraft and sorcery in the Gambia and Guinea-Bissau. The notion of ‘luck’ is understood here as an inherent quality of destiny, which is – paradoxically – pre-determined but contingent at the same time. Destiny is pre-set by God but requires human action and is also vulnerable to occult attacks. Such metaphysical interpretations produce (as well as express) specific attitudes to migration – hope and optimism on the one hand, fear and anxiety on the other. They also lead to certain strategies meant to avoid occult threats, which may be understood as ways of navigating uncertainty. I argue that it is the religious lens of destiny, ‘luck’ and, in a different way, witchcraft that is vital in dealing with the uncertainty of migration to the global North.

Performance Evaluation of M|M|1 Queueing Models in Cloud Computing Environments

Queues are ubiquitous, occurring at any given moment. Traffic congestion, healthcare facilities, dining establishments, and even the digital realm. With the continuous advancement of global society, there has been a significant surge in the utilization of the internet, leading to a substantial increase in the volume of incoming service requests occurring on a constant basis. The emergence of cloud computing can be attributed to its inherent advantages in terms of user comfort and cost-effectiveness. The utilization of cloud computing enables users to do computational tasks without the need for physical server ownership. However, it is important to note that the presence of queues is not eliminated as a result. Queueing theory was established as a means to address the issue of lengthy and sluggish lines. This research focuses on the analysis and optimization of queueing systems in order to develop efficient and expedient queues, while also considering cost-effectiveness. In the context of cloud computing, particular attention is given to the goal of energy conservation. This study aims to examine the performance of M|M| queues within the context of cloud computing. It will include both numerical and textual analyses, as well as proofs, while primarily focusing on comparing the similarities and differences of two M|M| queues.

Exploring online coalition building: A longitudinal analysis of interorganizational networks of nonprofit homeless shelters on social media

ABSTRACT While the networking nature of social media opens up opportunities for nonprofit e-advocacy, nonprofits’ coalition building through social media remains an understudied area of research. Drawing on previous research, we collected longitudinal Twitter data of nonprofit homeless shelters in the United States. By conducting a series of network and textual analyses, this study seeks to understand how social media’s connecting tools are used by shelter providers over time. We also observe the interorganizational networks that developed over time to examine whether online coalitions are built in the homelessness field. Results indicate that while nonprofit homeless shelters tend to underutilize the potential of social media for inter-organizational networking, the inter-organizational network tends to develop into a set of more complicated and stronger ties.

Neural natural language processing for long texts: A survey on classification and summarization

The adoption of Deep Neural Networks (DNNs) has greatly benefited Natural Language Processing (NLP) during the past decade. However, the demands of long document analysis are quite different from those of shorter texts, while the ever increasing size of documents uploaded online renders automated understanding of lengthy texts a critical issue. Relevant applications include automated Web mining, legal document review, medical records analysis, financial reports analysis, contract management, environmental impact assessment, news aggregation, etc. Despite the relatively recent development of efficient algorithms for analyzing long documents, practical tools in this field are currently flourishing. This article serves as an entry point into this dynamic domain and aims to achieve two objectives. First of all, it provides an introductory overview of the relevant neural building blocks, serving as a concise tutorial for the field. Secondly, it offers a brief examination of the current state-of-the-art in two key long document analysis tasks: document classification and document summarization. Sentiment analysis for long texts is also covered, since it is typically treated as a particular case of document classification. Consequently, this article presents an introductory exploration of document-level analysis, addressing the primary challenges, concerns, and existing solutions. Finally, it offers a concise definition of “long text/document”, presents an original overarching taxonomy of common deep neural methods for long document analysis and lists publicly available annotated datasets that can facilitate further research in this area.

AUTOMATED TEXT ANALYSIS OF HISTORICAL DOCUMENTS USING MACHINE LEARNING TECHNIQUES

This paper presents an innovative approach to the automated analysis of historical documents through the application of advanced machine learning techniques. The problem background in automated text analysis of historical documents using machine learning techniques is efficiently processing large volumes of diverse historical texts to extract valuable insights and patterns, enhancing historical research and understanding. Leveraging the power of natural language processing, this study proposes a comprehensive framework that encompasses data preprocessing, feature extraction, and model training, enabling the efficient extraction of valuable insights from vast collections of historical texts. By utilizing cutting-edge algorithms, including deep learning and sentiment analysis, the research demonstrates the potential of this approach in uncovering hidden patterns, sentiments, and semantic nuances within historical documents, thereby facilitating a deeper understanding of the past and shedding light on critical historical events and societal developments.