Latent Dirichlet Allocation Topic Model Research Articles

From policy to practice: decoding the evolution of K-12 English teacher training course selection in Zhejiang through topic modelling analysis (2014–2023)

ABSTRACT This article aimed to strengthen training programmes in Zhejiang by aligning them with educational policies to respond to ongoing reforms drawing on Latent Dirichlet allocation (LDA) topic modelling and content analysis. The topic modelling approach uncovers dominant training themes and trends by analysing extensive unstructured training topics from an online training platform in Zhejiang, thus bridging the gap between theoretical research and practical application. This approach offers data-driven insights, supporting dynamic tracking of educational trends in real-time for timely policy and curriculum adjustments. Ultimately, the article argues that real-time data and stakeholder feedback should be integrated to ensure educational policies align closely with actual teachers’ needs. However, due to data limitations, the article acknowledges a challenge in fully utilizing topic modelling and suggests combining it with qualitative analysis to gain a deeper understanding of educational practices. The Incorporation of a topic modelling approach into teacher education research could contribute significantly to training efficiency and policy relevance, fostering an adaptive educational policy environment.

Online review based IPA and IPCA: the case of Korean mobile banking apps

PurposeThis study introduces a novel approach to conducting importance-performance analysis (IPA) and importance-performance competitor analysis (IPCA) by utilizing online reviews as an alternative to traditional survey-based marketing research.Design/methodology/approachIn order to conduct IPA and IPCA utilizing online reviews, the following three steps were executed: (1) Extract key attributes of the product/service with latent Dirichlet allocation (LDA) topic modeling. (2) Measure the importance of each attribute with keyword analysis. (3) Measure the performance of each attribute with KoBERT-based sentiment analysis.FindingsBy utilizing LDA, we were able to identify significant attributes identified by real users’ reviews. The approach of evaluating attribute importance using keyword metrics offers the advantage of straightforward computations, reducing computational expenses and facilitating intuitive metric assessments. The evaluation of BERT-based performance involved adapting pre-trained language models to the specific analysis domain, resulting in substantial time savings without compromising accuracy, ultimately bolstering the dependability of the metrics. Lastly, the case study’s findings indicate a growing emphasis on the aesthetic aspects of mobile banking apps in South Korea while highlighting the pressing need for enhancements in critical areas such as stability and security, which are particularly pertinent to the finance industry.Originality/valuePrevious studies have limitations in assessing significance solely based on sentiment scores and review ratings, resulting in an inability to independently measure satisfaction and importance metrics. This research addresses these limitations by introducing a keyword frequency-based importance metric, enhancing the accuracy and suitability of these measurements independently. In the context of performance measurement, this study utilizes pre-trained large language models (LLMs), which provide a cost-effective alternative to previous methods while preserving measurement accuracy. Additionally, this approach demonstrates the potential for industry-wide competitive analysis by enabling comparisons among multiple competitors. Furthermore, the study extends the application of review data-based IPA and IPCA, traditionally used in the tourism sector, to the evaluation of financial mobile applications. This innovation expands the scope of these methodologies, indicating their potential applicability across various industries.

Text mining of syntactic complexity in L2 writing: an LDA topic modeling approach

Abstract The present study provides an overview of syntactic complexity (SC) in second language (L2) writing research, employing the Latent Dirichlet Allocation (LDA) topic modeling approach. Analyzing 470 abstracts sourced from the WoS Core Collection (2005–2023) and Scopus (1978–2023), this investigation explores the broad spectrum of research themes in SC in L2 writing. By employing perplexity and coherence tests, the study identifies the optimal number of topics in the model as five: writing quality, proficiency, genres, SC development, and task-based conditions. The finding reveals that: (1) In cross-sectional studies, the interplay among writing quality, proficiency levels, and genres often involves studying one variable as reliant on the others. Additionally, there is a growing inclination toward regarding phrasal structures as more dependable complexity indicators. (2) Longitudinal investigations predominantly emphasize the evolution of SC, yet there’s a scarcity of exploration in continuation writing tasks. (3) Regarding measurement metrics, a transition from broader SC evaluations to more intricate assessments is evident, though several studies lack thorough examinations, particularly in continuation tasks.

Advancing Social Equity in Urban UAV Logistics: Insights from the Academic Literature and Social Media

In recent years, the rapid growth of e-commerce and on-demand delivery services has placed a significant strain on urban logistics systems. Technological advances such as unmanned aerial vehicle (UAV)-based logistics systems have thus emerged as promising solutions in urban environments and are increasingly being piloted worldwide. However, the implementation of UAV logistics risks exacerbating social inequities, particularly in marginalized communities that may disproportionately bear the noise and safety risks. To mitigate these risks, it is crucial to integrate social equity considerations into urban UAV logistics. This study explores social equity factors through a systematic literature review and social media analysis of Xiaohongshu (the Little Red Book), a popular Chinese social media platform known for its extensive user base and active discussions on social issues. This literature review involves a full-text examination, while latent Dirichlet allocation (LDA) topic modeling is used to analyze social media comment datasets. Each method identifies social equity factors and separately assesses their relative importance, resulting in the final identification of 24 key factors that provide a holistic view of public sentiment and academic discourse. The findings reveal a divide between academic concerns around systemic risks and a public focus on immediate needs. By synthesizing these insights, this study provides a social equity landscape for urban UAV logistics and actionable references for policymakers and stakeholders.

AI-Driven Smart Transformation in Physical Education: Current Trends and Future Research Directions

Although the rapid development of Artificial Intelligence (AI) in recent years has brought increasing academic attention to the intelligent transformation of physical education, the core knowledge structure of this field, such as its primary research topics, has yet to be systematically explored. The LDA (latent Dirichlet allocation) topic model can identify latent themes in large-scale textual data, helping researchers extract key research directions and development trends from extensive literature. This study is based on data from the Web of Science Core Collection and employs a systematic literature screening process, utilizing the LDA topic model for in-depth analysis of relevant literature to reveal the current status and trends of AI technology in physical education. The findings indicate that AI applications in this field primarily focus on three areas: “AI and data-driven optimization of physical education and training”, “computer vision and AI-based movement behavior recognition and training optimization”, and “AI and virtual technology-driven innovation and assessment in physical education”. An in-depth analysis of existing research shows that the intelligentization of physical education, particularly in school and athletic training contexts, not only promotes sustainable development in the field but also significantly enhances teaching quality and safety, allowing educators to utilize data more precisely to optimize teaching strategies. However, current research remains relatively broad and lacks more precise and robust data support. Therefore, this study critically examines the limitations of current research in the field and proposes key research directions for further advancing the intelligent transformation of physical education, providing a solid theoretical framework and guidance for future research.

Topic modeling and content analysis of people’s anxiety-related concerns raised on a computer-mediated health platform

BackgroundAbout one in four Chinese people might suffer or have already suffered from anxiety conditions, with a lifetime prevalence rate of 4.8%. However, many of those who could have benefited from psychological or pharmacological treatments fail to be recognized or treated due to the lack of timely recognition and accurate diagnosis.ObjectiveThis study used a topic modeling approach to explore people’s anxiety-related concerns raised on a computer-mediated Chinese health platform, YOU WEN BI DA (questioning and answering), to provide implications for accurate diagnosis, targeted education, tailored intervention, and informed policy-making in the course of addressing this condition of public concern.MethodsFirst, we extracted data from YOU WEN BI DA between May 2022 and February 2023. After cleaning the extracted data both using the Python text processing tool spaCy and manually, we ascertained the optimal number of topics by drawing on the coherence scores and used latent Dirichlet allocation (LDA) topic modeling to generate the most salient topics and related terms. We then categorized the topics ascertained into different classes of themes by plotting them onto a 2D plane via multidimensional scaling using the pyLDAvis visualization tool. Finally, we analyzed these topics and themes qualitatively to better understand people’s anxiety-related concerns.Results5 topics with different overall prevalence were ascertained through data analysis. Topic 2 (tinnitus phobia-incurred concerns, n = 639) is the most popular dominant topic, occurring in 25.1% of the 2545 collected concerns, closely followed by Topics 1 (neurosis-incurred concerns, n = 512;) and 3 (sleep, dyskinesia, bipolar, cognitive, and somatic disorders-incurred concerns, n = 619), which appeared in 20.1% and 24.3% of the 2545 concerns respectively. Topic 5 (social phobia-incurred concerns, n = 428) ranks as the fourth most popular dominant topic, showing up in 16.8% of the 2545 concerns. Topic 4 (autonomic nerve dysfunction-incurred concerns, n = 347) accounts for 13.6% of the 2545 concerns. The t-distributed Stochastic Neighbor Embedding analysis reveals partial similarities between Topics 2 and 5 as well as between Topics 4 and 5 because many concerns involved in Topics 2 and 5 pertain to people’s psychological status of fear and anxiety and the relief and dispelling of such symptoms through medication, and many concerns involved in Topics 4 and 5 relate to people’s worries about the negative impact on their nerves and the adjustment and conditioning of such effects through medication.ConclusionThis was the first study that investigated Chinese people’s anxiety-related concerns raised on YOU WEN BI DA using the topic modeling technique. The automatic text analysis and complementary manual interpretation of the collected data allowed for the discovery of the dominant topics hidden in the data and the categorization of these topics into different themes to reveal the overall status of people’s anxiety-related concerns. The research findings can provide some practice implications for health and medical educators, practitioners, and policy-makers to make joint efforts to address this common public concern effectively and efficiently.

Dynamics of Sino-American relations: A decade-long analysis using topic modeling

Abstract. This study explores the thematic evolution of Sino-American relations over the past decade using Latent Dirichlet Allocation (LDA) topic modeling. We compiled a comprehensive dataset of academic papers, news articles, government reports, and social media posts, focusing on themes such as economic competition, technological rivalry, military strategy, diplomatic maneuvers, and ideological conflict. Our analysis identifies key sub-topics within these themes, including trade tariffs, 5G technology, and military expenditures, providing a nuanced understanding of the factors influencing bilateral relations. Temporal trends highlight the shifting prominence of these themes, with economic competition and technological rivalry becoming increasingly salient. Additionally, the concept of financial hegemony, particularly the role of the U.S. dollar, emerges as a significant driver of U.S. policy. This study offers insights into the strategic motivations behind U.S. actions and the complex interplay of various factors in Sino-American relations.

User Sentiment Analysis of the Shared Charging Service for China’s G318 Route

Shared charging services have gained popularity for their contribution to green travel. Accurately identifying the core factors that influence user experience (UX) not only enhances service quality and optimizes user satisfaction, but also promotes the dissemination of green travel concepts. However, the influencing factors and their mechanisms vary significantly across regions, particularly along the Chengdu–Lhasa (G318) route, which features large elevation changes, diverse climatic conditions, rugged terrain, and frequent geological disasters, making the influencing factors particularly complex. This study analyzes comment texts from 38 shared charging stations along the G318 route in the e-Charging APP, totaling 15,214 comments. A comprehensive approach is employed, including high-frequency word analysis, term frequency–inverse document frequency (TF-IDF) comparison, co-occurrence semantic network and co-word matrix feature correlation analysis, Latent Dirichlet Allocation (LDA) topic modeling, and sentiment analysis. This multifaceted analysis explores core themes, user viewpoints, and sentiments in the comments, focusing on users’ perspectives on service quality, usage experience, and environmental impact of the charging stations. The findings indicate that charging speed, service attitude, environment, operational status of hardware and software, and pricing are key factors influencing user sentiment. Users have a high demand for the perfection of supporting facilities of shared charging stations, directly affecting user satisfaction and indirectly influencing the brand image and market competitiveness of enterprises.

빅데이터 기반 지역사회개발 이슈 분석의 의미: 청도군 지역사회개발에 관한 LDA와 DMR 토픽모델링을 사례로

This study aims to discover the utility of big data analysis in research related to community development and to derive meaningful implications. To achieve this, the case of Cheongdo-gun, despite of a designated depopulation area, experiencing a rapid increase in temporary residents, was examined. Alongside a literature review on the region, Latent Dirichlet Allocation (LDA) and Dirichlet Multinomial Regression (DMR) topic modeling were conducted on 7,599 news articles from 89 media outlets collected from BigKinds between 2014 and 2023. This analysis enabled an understanding of the interrelations and changes among community development issues in Cheongdo-gun, such as cultural and artistic content, ecological and historical tourism, population influx policies, local festivals, and community activities. This study is significant in that it demonstrates how LDA and DMR topic modeling, using big data, objectively synthesizes specific issues within the sociocultural context of the community development.

Text analysis and optimization strategy of an app’s privacy policy from the perspective of dual perception: taking Chinese shopping apps as an example

Purpose This paper aims to conduct an in-depth analysis of the shortcomings of apps’ privacy policies and to propose improvement and optimization strategies, which are of great significance for establishing a transparent and responsible privacy protection framework that ensures compliant collection and use of users’ information and effective protection of their privacy. Design/methodology/approach This paper obtained privacy policy texts for 100 shopping apps through Web crawlers and manual downloads. Based on the perspective of perceived usefulness, thematic analysis is conducted through the latent Dirichlet allocation topic model and comparison with existing policies. Based on the perspective of perceived ease of use, readability analysis is conducted through content analysis and formula calculation. Findings The apps privacy policies can be divided into seven themes. The authors benchmark these seven topics with the Personal Information Protection Law of the People’s Republic of China, the E-Commerce Law of the People’s Republic of China and the General Data Protection Regulation. It is found that there are omissions in the information collection and use and juvenile protection of the existing apps. Through the indicators’ readability analysis and calculation, it is found that the existing apps privacy policies have good performance in the readability indicators such as naming method, frame directory and so on. However, text personalization and text readability need to be improved and optimized. Originality/value At the theoretical level, this paper constructs a model from the dual perception perspectives of perceived usefulness and perceived ease of use and analyses the apps’ privacy policy texts at a fine-grained level. At the practical level, based on large-scale apps’ privacy policy text data, this paper conducts multi-dimensional research from theme analysis, authoritative law benchmarking analysis, content analysis and text readability calculation and analysis. At the same time, this paper identifies the current problems of apps’ privacy policies and puts forward countermeasure suggestions for their content improvement and optimization.

Reducing Consumer–Brand Incongruity Through Corporate Social Responsibility and Brand Trust: Exploring Negative Word‐of‐Mouth (NWOM)

ABSTRACTDrawing upon consumer–brand disidentification theory and balance theory, this study examines symbolic and ideological incongruity in consumer–brand relationships through an original conceptual model shaped by negative past experiences, brand trust, perceived corporate social responsibility (CSR), and negative word‐of‐mouth (NWOM). A preliminary study was conducted to explore the dimensions of consumers' negative past experiences by topic detection. Latent Dirichlet allocation (LDA) topic modeling was undertaken to analyze online consumer reviews (n = 6095) about a coffee chain brand. The dimensions detected in this preliminary study were included in the research model and further analyzed in the main study. The main study, a cross‐sectional consumer survey (n = 522), tested the original research model by way of partial least squares structural equation modeling (PLS‐SEM) on SmartPLS. The findings showed that negative past experiences consisted of product‐related, service‐related, and technology‐related problems and negatively influenced brand trust. It was found that brand trust and perceived CSR negatively affected symbolic and ideological incongruity, while symbolic and ideological incongruity positively influenced NWOM. The findings provide empirical evidence for balance theory by showing that the three critical domains of consumer–brand relationships (ideological, symbolic, and experiential) provide a complex cognitive model that covers personal‐symbolic and moral‐societal aspects of consumer–brand disidentification and consequent NWOM intentions. In line with consumer–brand disidentification theory, the results contribute to the literature by demonstrating the direct negative impacts of brand trust and perceived CSR on symbolic and ideological incongruity, as well as the direct positive impacts of symbolic and ideological incongruity on NWOM.

Leveraging user’s preference and social circle for personalized recommendation via matrix factorization with sub-linear convergence rate

Recommender systems aim to filter information effectively and recommend useful sources to match users’ requirements. However, the exponential growth of information in recent social networks may cause low prediction accuracy for recommendation systems. This article proposes a unified personalized recommendation architecture referred to as PSRec, which incorporates user preference and social relationship into matrix factorization framework. Specifically, PSRec generates two collections for textual reviews and contextual information respectively, and performs preference learning for each user via the Latent Dirichlet Allocation topic model. Moreover, PSRec exploits the inner relations within the social circle for recommendation, including direct trust relationship and indirect trust relationship. Additionally, it’s certificated that PSRec can converge at a sub-linear rate via theoretical analysis. Experimental results over DoubanMovie, CiaoDVDs and Yelp demonstrate the superiority of the proposed PSRec, which can achieve significant improvements and provide much better user experience while compared with other benchmark models.

Corporate innovation and performance: a textual measure based on securities analysts’ research reports

ABSTRACT The relationship between innovation investment and firm performance has been a central focus of academic research. Traditional metrics, such as patent counts and R&D expenditures, are often criticised for their limited scope and lack of diversity. In contrast, analyst reports offer a richer source of information on innovation activities. This study employs textual analysis to assess firms’ innovation efforts by examining research reports on publicly listed companies. Using methods such as word segmentation, the research constructs corporate innovation indicators based on the LDA (Latent Dirichlet Allocation) topic model. Analyzing data from 2010 to 2021, the study empirically validates these new innovation indicators and investigates their impact on firm performance. The findings indicate that corporate innovation significantly enhances firm performance. Compared to traditional metrics like patents, the indicators developed here are more comprehensive, offering a broader measure of innovation and addressing the limitations of conventional metrics. These results are consistent across various robustness checks. Additionally, the study finds that corporate innovation improves performance primarily by increasing total factor productivity. Furthermore, strong corporate governance amplifies the positive impact of innovation on firm performance.

Exploring a Language-Based Interest Assessment: Predicting Vocational Interests on Social Media Using Natural Language Processing

For more than a century, self-report inventories have been the traditional method for assessing vocational interests. Little research has examined the use of machine learning techniques, such as natural language processing (NLP), in interest assessment. This paper explores the extent to which natural language on social media can be used to predict individuals’ self-ratings on eight basic interests representing the SETPOINT model: Agriculture, Engineering, Human Resources, Life Science, Management/Administration, Mechanics/Electronics, Media, and Social Science. Leveraging closed- (Linguistic Inquiry and Word Count; LIWC) and open-vocabulary NLP approaches (Latent Dirichlet allocation (LDA) topic modeling), we analyzed 3.2 million Facebook posts from 2,834 participants who completed a 32-item basic interest measure. We found that the convergent validities of these NLP approaches for predicting vocational interest scores (LIWC: [Formula: see text] = 0.19; LDA topic modeling: [Formula: see text] = 0.24) are comparable to prior research on language-based personality assessments. Our study also revealed largely face-valid language markers that characterize different basic interests. Implications for developing language-based interest assessments for applied settings (e.g., career guidance and employee selection) and future research directions are discussed.

Topic selection for text classification using ensemble topic modeling with grouping, scoring, and modeling approach

TextNetTopics (Yousef et al. in Front Genet 13:893378, 2022. https://doi.org/10.3389/fgene.2022.893378) is a recently developed approach that performs text classification-based topics (a topic is a group of terms or words) extracted from a Latent Dirichlet Allocation topic modeling as features rather than individual words. Following this approach enables TextNetTopics to fulfill dimensionality reduction while preserving and embedding more thematic and semantic information into the text document representations. In this article, we introduced a novel approach, the Ensemble Topic Model for Topic Selection (ENTM-TS), an advancement of TextNetTopics. ENTM-TS integrates multiple topic models using the Grouping, Scoring, and Modeling approach, thereby mitigating the performance variability introduced by employing individual topic modeling methods within TextNetTopics. Additionally, we performed a thorough comparative study to evaluate TextNetTopics’ performance using eleven state-of-the-art topic modeling algorithms. We used the extracted topics for each as input to the G component in the TextNetTopics tool to select the most compelling topic model regarding their predictive behavior for text classification. We conducted our comprehensive evaluation utilizing the Drug-Induced Liver Injury textual dataset from the CAMDA community and the WOS-5736 dataset. The experimental results show that the Latent Semantic Indexing provides comparable performance measures with fewer discriminative features when compared with other topic modeling methods. Moreover, our evaluation reveals that the performance of ENTM-TS surpasses or aligns with the optimal outcomes obtained from individual topic models across the two datasets, establishing it as a robust and effective enhancement in text classification tasks.

Understanding privacy concerns in ChatGPT: A data-driven approach with LDA topic modeling

This study investigates privacy concerns associated with ChatGPT, a prominent generative AI model, through a data-driven approach combining Twitter data analysis and a user survey. Leveraging Latent Dirichlet Allocation (LDA) topic modeling and data categorization techniques, the research identifies key areas of concern: 1) Privacy Leakage Due to Public Data Exploitation, 2) Privacy Leakage Due to Personal Input Exploitation, and 3) Privacy Leakage Due to Unauthorized Access. Twitter data analysis of over 500k tweets, supplemented by a survey of 67 ChatGPT users, reveals nuanced user perceptions and experiences regarding privacy risks. A Python program was used to improve a dataset of 500k tweets referencing “ChatGPT” during the data preparation stage. To get a refined collection of terms, steps included converting text to lowercase, eliminating mentions and hyperlinks, tokenizing, eliminating stopwords, and keyword matching to extract tweets about ChatGPT's privacy features. Once preprocessing was completed, there were 11k refined tweets. Results highlight significant apprehensions, particularly regarding unauthorized access, underscoring the importance of robust privacy measures in AI systems. The study contributes to understanding user concerns, informing policy decisions, and guiding future research on privacy in generative AI. These studies might improve ChatGPT and other AI systems' security and privacy. The public, corporations, researchers, lawmakers, and AI developers may all benefit from the useful information it provides in better understanding and managing privacy threats.

토픽 모델링을 통한 대학 예비교수자가 인식하는 좋은 수업의 특성 탐색

The study aims to identify the core keywords of ‘good instruction’ as perceived by prospective professors and to explore the the characteristics of ‘good instruction’. A total of 142 teaching philosophy essays were collected from the prospective professors and analyzed using the Latent Dirichlet Allocation (LDA) topic modeling method, resulting in the identification of five main topics. First, the core keywords frequently appearing in good instruction included ‘people’, ‘questions’, and ‘preparation’. Second, the characteristics of good instruction were categorized into five topics: encouraging learner participation through diverse methods, promoting academic achievement through continuous monitoring of learning, supporting the development of expertise in the major field, providing a safe learning environment based on mutual trust, and fostering active communication and interaction. This study is significant as it provides a framework and direction for what constitutes good instruction in higher education based on the teaching philosophies of prospective professors. It also offers practical implications for design and implementation of good instruction.

How to go green? Exploring public attention and sentiment towards waste sorting behaviors on Weibo platform: A study based on text co-occurrence networks and deep learning

The attention and sentiment of the public are crucial for better implementation of waste sorting behaviors and moving towards green living. In this study, web scraping technology was used to collect 367,856 Weibo posts related to waste sorting from Sina Weibo. Utilizing text co-occurrence networks, Latent Dirichlet Allocation (LDA) topic modeling, and a deep learning model combining the Affective Cognitive Model (OCC) with Long Short-Term Memory Model (LSTM) (referred to as OCC-LSTM), we comprehensively understand the text at both micro and macro levels, analyzing the attention and sentiment of the public towards waste sorting behaviors on the Sina Weibo platform. Several important findings emerged from the empirical results. First, highly engaging posts were predominantly published by users with a large following, and the number of posts fluctuated over time. This reflects the influence of social hot topics and the timeliness of information dissemination. Second, there was heterogeneity in the user groups and their locations, often influenced by cultural differences due to geographical location. Third, positive sentiment towards waste sorting behavior was higher than negative sentiment on the Weibo platform. Moreover, public attention varied under different emotional influences concerning the topic of waste sorting behavior. The innovation of this study lies in the development of a research framework combining co-occurrence networks and deep learning, expanding the analysis on both micro and macro levels. This framework broadens the research paradigms and dimensions of public perception in waste sorting. This study is significant for promoting waste sorting behaviors and implementing climate policies.

Fighting for collusive bidding in the construction industry: A text mining-enabled approach

Policy measures are crucial for regulating collusive bidding and are integral to effective governance. However, current research lacks a comparative exploration of strategies to combat collusive bidding through policy. Therefore, this study aims to identify more effective countermeasures by examining policy variations between regions with low and high incidences of collusive bidding. Using Latent Dirichlet Allocation (LDA) topic modeling, the study extracts key themes from these policies, while qualitative analysis highlights differences in approaches. It underscores that integrating electronic and informatization technologies into bidding systems significantly reduces collusive practices. While increasing penalties can deter collusive bidding, achieving desired impacts requires thorough investigation and vigilant oversight. Additionally, strengthening external supervision enhances control over such activities. This study identifies critical governance strategies for addressing collusive bidding and advocates further research into more effective methods within the construction sector.

Unveiling tourist typology: using online reviews and LDA to understand motivations for visiting Kyoto's prominent temples

Purpose This study aims to identify motivations for visiting Kyoto's prominent religious attractions using latent Dirichlet allocation (LDA) text analysis of online reviews; establish linkages between push motivational factors and pull factors of the religious sites, forming distinct tourist typologies; and suggest strategies for Kyoto's destination marketing based on the findings. Design/methodology/approach This study analyzed 37,772 TripAdvisor reviews for Kyoto's top 25 religious sites from the pre-pandemic period (March 2020). LDA topic modeling extracts 18 underlying thematic dimensions from the review texts. Axial coding of these dimensions revealed five distinct tourist motivation typologies. Findings Five motivation typologies emerged: cultural seekers drawn to Japan's unique heritage, nature lovers attracted by scenic landscapes, chrono-seasonal experiencers seeking distinct seasonal views, crowd-avoiders prioritizing less congested visits and city wanderers engaging in local activities. Practical implications The findings offer valuable guidance for destination marketers and managers in Kyoto, enabling the development of targeted strategies to enhance visitor experiences and manage overcrowding at popular religious sites. Originality/value This research provides novel insights into nonreligious tourists' motivations for visiting religious sites in a crowded destination. By identifying distinct motivation-based tourist typologies, the study informs strategies for enhancing visitor experiences tailored to diverse needs, contributing to tourism literature and practical destination management.