Structural Topic Model Research Articles

Comparison of patenting trends and alliance patterns among countries with a majority religion: structural topic model and network analysis

ABSTRACT In our modern, technologically advanced society, 84% of the world’s population holds some form of religious faith. Our history has also witnessed religion and technology having an impact on each other. Assuming the underlying correlation between them, we question whether there are distinguished technological features among countries with strong religions. This study examines the technological trends in countries with a majority religion according to the dominant religion of each – namely, Judaism, Islam, and Hinduism. We applied structural topic modelling (STM) and co-patent network analysis to patents from those countries registered with the United States Patent and Trademark Office between 1965 and 2015. According to the STM results, technologies appear to reflect geographical environments, industrial concentrations, and investments led by governments and private sectors rather than religion. Moreover, countries with a majority religion in our study had mainly collaborated with the United States and European countries, regardless of religious tendency.

Speaking of civilians: Automated text analysis of the United Nations’ framing of complex humanitarian emergencies

Since the end of the Cold War, the protection of civilians has increased its weight on the United Nations (UN) agenda. This article (1) maps the evolution of civilian protection within the UN framework as an indicator of its shifting priorities, (2) identifies breakpoints in the prevalence and character of the protection discourse, and (3) explores how internal processes of policy development and real-world triggers (namely conflicts and peacekeeping operations) shaped this transformation. The article uses Structural Topic Modeling (STM) to analyze an original corpus of Security Council and General Assembly resolutions on complex humanitarian emergencies since 1990. The analysis uncovers two distinct forms of protection, labeled as “Ground protection” and “Political–legal protection,” which are characterized by contrasting temporal and geographic trajectories. Moreover, critical junctures in the protection rhetoric (during the years 2000, 2005, and 2008) coincide with policy watersheds rather than conflict outbreaks or trends in peacekeeping deployment. This article offers a comprehensive analysis of the intricate evolution of civilian protection using text-as-data methods, which uses its theory-building design to encourage further explorations on the interplay between internal and external factors in shaping its progression within the UN framework.

Analysis adapted from text mining quantitively reveals abrupt and gradual plant-community transitions after fire in sagebrush steppe

ContextPlant communities vary both abruptly and gradually over time but differentiating between types of change can be difficult with existing classification and ordination methods. Structural topic modeling (STRUTMO), a text mining analysis, offers a flexible methodology for analyzing both types of temporal trends.ObjectivesOur objectives were to (1) identify post-fire dominant sagebrush steppe plant association types and ask how they vary with time at a landscape (multi-fire) scale and (2) ask how often major association changes are apparent at the plot-level scale.MethodsWe used STRUTMO and plant species cover collected between 2002–2022 across six large burn areas (1941 plots) in the Great Basin, USA to characterize landscape change in dominant plant association up to 14 years post-fire. In a case study, we assessed frequency of large annual changes (≥ 10% increase in one association and decrease in another) between associations at the plot-level scale.ResultsSTRUTMO revealed 10 association types dominated by either perennial bunchgrasses, mixed perennial or annual grasses and forbs, or exotic annual grasses. Across all study fires, associations dominated by large-statured perennial bunchgrasses increased then stabilized, replacing the Sandberg bluegrass (Poa secunda)-dominated association. The cheatgrass (Bromus tectorum)-dominant association decreased and then increased. At the plot-level, bidirectional changes among associations occurred in ~ 75% of observations, and transitions from annual invaded to perennial associations were more common than the reverse.ConclusionsThe analysis revealed that associations dominated by some species (i.e. crested wheatgrass, Agropyron cristatum, Siberian wheatgrass, Agropyron fridgida, or medusahead, Taeniatherum caput-medusae) were more stable than associations dominated by others (i.e. Sandberg bluegrass or cheatgrass). Strong threshold-like transitions were not observed at the multi-fire scale, despite frequent ephemeral plot-level changes.

교육과정 개정과 대입제도 개편에 따른 계열 간 교차지원 관련 언론 보도 텍스트 분석

The purpose of this study is to analyze the impact of national curriculum reform and college admissions reform on the evolving nature of cross-disciplinary college applications and to identify the key issues associated with cross-disciplinary college applications through an analysis of news articles published by mainstream mass media outlets. For this purpose, a total of 2,371 news articles related to cross-disciplinary college applications published by mainstream mass media outlets from January 1, 1997, to Novermber 30, 2023, were collected using web scraping method, and keyword frequency analysis and Structural Topic Modeling (STM) were conducted. The main results of the study are as follows. First, over the course of 26 years, 7 topics related to cross-disciplinary college applications reported in news articles were extracted, including ‘Plans for the promotion of public education and reduction of private education expenses’, ‘Policies for supporting students with specialized talents’, ‘Early admissions procedures and support strategies’, ‘Trends of natural science students enrolling in liberal arts majors and the introduction of integrated CSAT’, ‘Avoidance of science and engineering majors and support policies’, ‘Application strategies based on admission types’, and ‘Trends of liberal arts students enrolling in natural science majors and advantages and disadvantages based on admission types and subjects’. Second, the appearance rate of topics and the major issues varied across different phases of the integrated education system. Lastly, based on the results of this study, implications regarding the issues and directions for improvement of support were derived.

Thinking different as an act of resistance: Reconceptualizing the German protests in the COVID-19 pandemic as an emergent counter-knowledge order

Massive anti-government protests erupted during the COVID-19 pandemic in Germany. The crisis activated a potential for resistance that has been simmering under the impositions of late-modern knowledge society. Made salient by the pandemic conditions of sudden extreme reliance on scientific (non) knowledge, the corona protestors activated this potential for resistance and constructed their own counter-knowledge order bound by shared resentment of and distrust in the established order and facilitated by digital platforms. Utilising social network analysis and structural topic modeling for digital critical discourse analysis, in this paper I explore how the corona protest counter-knowledge order is constructed with a particular focus on its contexts, roles, and hierarchies. I find that far-right and conspiracy imaginations are used to level out hierarchies and detach epistemic roles from their contexts to reinstate a superior self into interpretative power. The counter-knowledge order’s inherent construction of unwarranted omnipotence points to a more fundamental resistance to the established normative orders of our society that should be addressed more effectively if we want to be prepared for future crises and not lose common ground for making sense of them.

Investigating the Coverage of China’s Vaccine Crisis on YouTube: Networked Framing, Grassroots Activism, and Homophily

ABSTRACT This study investigates how China’s domestic vaccine crisis in 2018 was politicized within Chinese video content on YouTube. The analysis, employing the Structural Topic Model, uncovered a notable hyper-politicization of the vaccine crisis in video comments, often diverging from the issue frames presented in video titles. Furthermore, inferential network analysis using the Exponential Random Graph Model revealed the hyper-politicized comments can be attributed to 1) the emergence of networked commenters engaging in cross-video commenting practices, and 2) political homophily among video producers, particularly regarding their critical political stance toward China. The findings suggest the networked dynamics of ssue politicization.

Political Discourse on Twitter: The Case of Election Fraud in South Korea

The fake news that circulated on social media in 2020 and 2022 affected the voters of the South Korean general and presidential elections, respectively. The political controversy, centered on the election fraud that occurred in both the elections, received significant attention from the society. This study emphasizes the Twitter discourse and compares the formation and distribution of election fraud. We conduct semantic social network analysis and structural topic modeling (STM) to represent topics and relationships among the emerging themes using Twitter texts related to both elections. Results demonstrate that discourses on the same political issue exhibited unique contents and structures, as information is formed and distributed differently depending on the type of election in South Korea. Furthermore, this study illustrates the process and analysis of large-scale text data collected from Twitter. It also includes new methods, such as political ideology, for considering an analytical dimension.

How local governments balance multiple agendas in small-scale outbreaks responses: A multi-case comparative study from four major cities in China

Following the initial COVID-19 pandemic in China, major cities faced the critical task of balancing pandemic prevention and control with economic recovery. This study explores how each local government allocates its attention resources by analyzing textual information related to COVID-19 in the official websites of Beijing, Dalian, Qingdao, and Shanghai. Using Structural Topic Modeling (STM) and Social Network Analysis (SNA), we identified the main features of textual topics, the relationships between topics, the trends over time, and how the topics changed across document types. Our analysis showed that local governments set priorities based on their administrative autonomy and the severity of the epidemic. Therefore, resilience building in cities against epidemics should consider both factors and adopt differentiated strategies and multilevel approaches.

Structural topic modeling-based systematic review for service robots’ fault diagnosis: content, popularity and regional differences

Abstract The market for service robots is expanding as labor costs continue to rise. Faced with intricate working environments, fault detection and diagnosis are crucial to ensure the proper functioning of service robots. The objective of this review is to systematically investigate the realm of service robots’ fault diagnosis through the application of Structural Topic Modeling. A total of 289 papers were included, culminating in ten topics, including advanced algorithm application, data learning-based evaluation, automated equipment maintenance, actuator diagnosis for manipulator, non-parametric method, distributed diagnosis in multi-agent systems, signal-based anomaly analysis, integrating complex control framework, event knowledge assistance, mobile robot particle filtering method. These topics spanned service robot hardware and software failures, diverse service robot systems, and a range of advanced algorithms for fault detection in service robots. Asia-Pacific, Europe, and the Americas, recognized as three pivotal regions propelling the advancement of service robots, were employed as covariates in this review to investigate regional disparities. The review found that current research tends to favor the use of artificial intelligence (AI) algorithms to address service robots’ complex system faults and vast volumes of data. The topics of algorithms, data learning, automated maintenance, and signal analysis are advancing with the support of AI, gaining increasing popularity as a burgeoning trend. Additionally, variations in research focus across different regions were found. The Asia-Pacific region tends to prioritize algorithm-related studies, while Europe and the Americas show a greater emphasis on robot safety issues. The integration of diverse technologies holds the potential to bring forth new opportunities for future service robot fault diagnosis.Simultaneously, regional standards about data, communication, and other aspects can streamline the development of methods for service robots’ fault diagnosis.

“Letter to my future self” as a device for assessing health education effectiveness

ObjectiveTo examine whether a “letter to my future self” analyzed using structural topic modeling (STM) represents a useful technique in revealing how participants integrate educational content into planned future behaviors. Methods453 club-sports athletes in a concussion-education randomized control study wrote two-paragraph letters describing what they hoped to remember after viewing one of three randomly assigned educational interventions. ResultsA six-topic solution revealed three topics related to the content of the education and three topics related to the participant behavioral takeaways. The content-related topics reflected the educational content viewed. The behavioral takeaway topics indicated that the Consequence-based education was more likely to generate the Concussion Seriousness[CS:23%] topic while Traditional(24%) and Consequence-based(20%) interventions were more likely to generate the Responsibility for Brain Health[BH] topic. Traditional(21%) and Revised-symptom(17%) interventions were more likely to generate the Awareness and Action topics. ConclusionUnstructured user-generated data in the form of a “letter to my future self” analyzed using structural topic modeling provides a novel evaluation of the present and likely future impact of educational interventions. Practice implicationsPatient educators can enhance the effectiveness of education through the application of these methods to the evaluation of and innovation in programs.

Is industrial policy back in fashion? Text-as-data evidence from UK policy documents

Abstract This article evaluates the claim that industrial policy is seeing a revival in developed economies, using text-as-data evidence from UK government policy papers. Structural topic modeling shows that content which can be related to industrial policy has indeed seen a large increase in prevalence over the past decade compared to the baseline of the post-1980 liberal era. Moreover, such content is shown to be increasingly central to post-2010 economic policy based on its position in the network of topics, on the number of downloads of documents associated with it, and on inclusion in important papers. An automated text summarization algorithm is used to extract the fragments which are most representative for these developments, and these are shown to closely match common definitions of industrial policy. A sentiment analysis algorithm is then used to extract the motivations given for policy proposals in representative documents, and indicates that declining economic competitiveness is a central concern.

Gender roles, perspectives, and issue attention in the Italian political twitterverse. An analysis of politicians’ network and top-down communication

ABSTRACT We investigate gender disparities in politics through an exploration of the Italian political Twitterverse. Using network analysis and structural topic models on a novel dataset of Italian politicians’ tweets (January 2020 - February 2022), we find that women are as well-connected to their political peers and as active as their male colleagues, and tweet on a wide range of topics emphasizing the sub-issues they care the most. These findings indicate that, as for women’s positioning within the politicians’ network, their activism and issue attention, Twitter seems to be a more favorable environment than more conventional arenas, where numeric under-representation, appointment to marginal positions, and gender gaps in issue attention are still detectable.

Examining school sector and mission in a landscape of parental choice

Researchers have considered how school choice policies affect student achievement, but less inquiry explores how the organization of schools may change in the presence of choice. This descriptive and exploratory paper analyzes a state representative sample of school mission statements at two time points: before the enactment of choice policies in Indiana, namely the Indiana Choice Scholarship Program, and again six years into the policy. Using structural topic modeling, this paper examines whether and how school mission statements topics have changed over this period. Descriptive findings suggest mission statement topics differ significantly between sectors but show few changes over time. The most striking shift is that Catholic and other private religious schools appear to be clarifying the religious aspects of their mission in the presence of robust choice policies.

Discovering latent themes in aviation safety reports using text mining and network analytics

Aviation accidents, referring to unexpected and undesirable events involving aircraft, often cause great damage to property and human life. Learning from historical accidents is pivotal for improving safety in aviation. However, aviation accidents are typically documented and stored as unstructured or semi-structured free-text, rendering the ability to analyze such data a difficult task. This study presents a novel framework that combines text mining and network analytics techniques and provides the ability to analyze aviation accident reports automatically. The framework comprises a four-step modelling approach to: (1) the transformation of unstructured aviation safety report texts into structured numeric matrices using the TF-IDF matrix; (2) the identification of aviation accident topics using a Structural Topic Model (STM); (3) the production of a Word Co-occurrence Network (WCN) to determine the interrelations between aviation safety risk factors; and (4) quantitative analysis by technology of keywords to pinpoint key causal factors in aviation safety events. The proposed framework is validated by analyzing aviation accident reports collected by the National Transportation Safety Board (NTSB). The results indicate that STM provides a more granular partitioning of topics and better distinguishes between similar events compared to traditional Latent Dirichlet Allocation (LDA). Among the identified topics, “Fuel and Power” and “En-route Phase” have the highest occurrence rate according to STM. Additionally, “Aircraft Crash” is the most prevalent topic in aviation accidents that resulted in fatal injuries, while “Landing phase” is the most prevalent topic in non-fatal injuries on accidents. Based on the word co-occurrence network, three centrality measures highlight “inspection of equipment” and “take off” as the most important risk factors in aviation safety. The proposed framework provides a comprehensive solution for in-depth analysis of aviation safety reports, offering decision support for aviation safety management and accident prevention, thereby reducing risks and strengthening safety measures.

Reasons Behind Words: OPEC Narratives and the Oil Market

We analyze the content of the Organization of the Petroleum Exporting Countries(OPEC) communications and whether it provides information to the crude oil market. To this end, we derive an empirical strategy which allows us to measure OPEC’s public signal and test whether market participants find it credible. Using Structural Topic Models, we analyze OPEC narratives and identify several topics related to fundamental factors, such as demand, supply, and speculative activity in the crude oil market. Importantly, we find that OPEC communication reduces oil price volatility and prompts market participants to rebalance their positions. Our analysis indicates that market participants assess OPEC communications as providing an important signal to the crude oil market.

Semantic Networks of Election Fraud: Comparing the Twitter Discourses of the U.S. and Korean Presidential Elections

Traditional news outlets, such as newspapers and television, are no longer major sources of news. These media channels have been replaced by social platforms, which have increased in value as information distributors. This change in communication is an underlying reason for the election fraud controversies that occurred in the United States and South Korea, which hold high standards of democracy, during similar periods. This study investigates a model for sharing political disputes over social networks, especially Twitter, and illustrates the influence of political polarization. This study examines Twitter content around the presidential elections in the United States and South Korea in 2020 and 2022, respectively. It applies semantic network analysis and structural topic modeling to describe and compare the dynamics of online discourse on the issue of election fraud. The results show that online spaces such as Twitter serve as public spheres for discussion among active political participants. Social networks are key settings for forming and spreading election fraud controversies in the United States and South Korea, with differences in content. In addition, the study applies large-volume text data and new analytical methods such as the structural topic model to examine the in-depth relationships among political issues in cyberspace.

구조적 토픽 모형을 활용한 기초학력 관련 담론 및 정책 동향 분석

Purpose: This study aims to investigate discourse and policy trends in basic academic achievement
 by analyzing texts associated with the subject. Exploring main topics and their changes in response
 to the dynamics of basic academic achievement guarantee policies will provide insights that can
 contribute to the development and improvement of relating policies.
 Methods: Structural Topic Modeling (STM) was employed to extract topics in analyzing academic
 journal articles, Ministry of Education press releases, and media articles. The topic prevalence was
 estimated and statistically tested in relation to the phases of the basic academic achievement guarantee
 policy, namely growth period, transition period, and confusion period. These policy phases served
 as metadata for the analysis.
 Results: Seven topics were extracted from academic journal articles, including academic achievement
 assessment, the characteristics of underachievers, and basic academic achievement policies. Another
 seven topics emerged in the Ministry of Education press releases, including basic academic achievement
 improvement programs, customized support for educational welfare, and online class and learning loss
 support. In media articles, nine topics were identified, including basic academic achievement
 improvement programs, support for linking specialized institutions, and controversies over how to
 conduct diagnostic evaluation. The prevalence of these topics changed over the policy phases.
 Conclusion: The study suggests that research on learner characteristics and vulnerable groups should
 be continued and expanded. It also emphasizes the importance of establishing long-term policies,
 fostering social consensus to strengthen the diagnosis of basic academic achievement, and providing
 support to enhance teachers’ expertise on basic academic achievement.

Structural topic modeling for corporate social responsibility of food supply chain management: evidence from FDA recalls on plant-based food products

PurposeThe rising number of food recalls has raised concerns about complexity, globalization and weak governance in the food supply chain. This paper aims to investigate the recall of plant-based products with data from the US Food and Drug Administration.Design/methodology/approachIntroducing the structural topic modeling method allowed us to test theories on recall in the context of sustainable food consumption, enhancing the understanding of food recall processes. This approach helps identify latent topics of product recalls and their interwoven relationships with various stakeholders.FindingsThe results answer a standing research call for empirical investigation in a nascent food industry to identify stakeholders’ engagements for food safety crisis management for corporate social responsibility practices. This finding provides novel insights on managing threats to food safety at an industry level to extend existing antecedents and consequences of product recall at a micro level.Practical implicationsFor practitioners, this empirical finding may provide insights into stakeholder management and develop evidence-based strategies to prevent threats to food safety. For public policymakers, this analysis may help identify patterns of recalls and assist guidelines and alarm systems (e.g. EU’s Rapid Alert System for Food and Feed) on threats in the food supply chain.Originality/valueTwo detected clusters, such as opportunisms of market actors in the plant-based food system and food culture, from the analysis help understand corporate social responsibility and food safety in the plant-based food industry.

Structural topic models for the automated measurement of entrepreneurial motivation among college students of different disciplines

ABSTRACT Qualitative methods can measure student motivation and evaluate entrepreneurship education (EE) with greater refinement than what is possible with purely quantitative methods, but impose labor-intensive labeling. Machine learning (ML) can automate qualitative analysis and supports in-depth exploration (versus mere description) of large qualitative datasets, but requires adaptation to the application domain, and data scientists have rarely applied ML to EE. In this article, we apply the structural topic model (STM), a type of natural language processing, to measure the entrepreneurial motivation of students. Once adapted to EE, STM finds that students belonging to two different colleges (computer science and management) in a U.S. polytechnic university are motivated by different topics, including environment, leadership, teamwork, strategy, social personal values, passion, and opportunity. This article endeavors to bridge the gap between EE research and ML, and stimulate the adoption of emerging qualitative ML techniques.

Disentangling boredom from depression using the phenomenology and content of involuntary autobiographical memories

Recurrent involuntary autobiographical memories (IAMs) are memories retrieved unintentionally and repetitively. We examined whether the phenomenology and content of recurrent IAMs could differentiate boredom and depression, both of which are characterized by affective dysregulation and spontaneous thought. Participants (n = 2484) described their most frequent IAM and rated its phenomenological properties (e.g., valence). Structural topic modeling, a method of unsupervised machine learning, identified coherent content within the described memories. Boredom proneness was positively correlated with depressive symptoms, and both boredom proneness and depressive symptoms were correlated with more negative recurrent IAMs. Boredom proneness predicted less vivid recurrent IAMs, whereas depressive symptoms predicted more vivid, negative, and emotionally intense ones. Memory content also diverged: topics such as relationship conflicts were positively predicted by depressive symptoms, but negatively predicted by boredom proneness. Phenomenology and content in recurrent IAMs can effectively disambiguate boredom proneness from depressive symptoms in a large sample of undergraduate students from a racially diverse university.