Résultats de recherche pour « Brand-Topic Model »

This article highlights how social media data and language analysis can help managers understand brand positioning and brand competitive spaces to enable them to make various strategic and tactical decisions about brands. The authors use the output of topic models at the brand level to evaluate similarities between brands and to identify potential cobrand partners. In addition to using average topic probabilities to assess brands’ relationships to each other, they incorporate a differential language analysis framework, which implements scientific inference with multi-test-corrected hypothesis testing, to evaluate positive and negative topic correlates of brand names. The authors highlight the various applications of these approaches in decision making for brand management, including the assessment of brand positioning and future cobranding partnerships, design of marketing communication, identification of new product introductions, and identification of potential negative brand associations that can pose a threat to a brand's image. Moreover, they introduce a new metric, “temporal topic variability,” that can serve as an early warning of future changes in consumer preference. The authors evaluate social media analytic contributions against offline survey data. They demonstrate their approach with a sample of 193 brands, representing a broad set of categories, and discuss its implications.

Topic Modeling is a well-known unsupervised learning technique used when dealing with text data. It is used to discover latent patterns, called topics, in a collection of documents (corpus). This technique provides a convenient way to retrieve information from unclassified and unstructured text. Topic Modeling tasks have been performed for tracking events/topics/trends in different domains such as academic, public health, marketing, news, and so on. In this paper, we propose a framework for extracting topics from a large dataset of short messages, for brand interest tracking purposes. The framework consists training LDA topic models for each brand using time intervals, and then applying the model on aggregated documents. Additionally, we present a set of preprocessing tasks that helped to improve the topic models and the corresponding outputs. The experiments demonstrate that topic modeling can successfully track people’s discussions on Social Networks even in massive datasets, and ca pture those topics spiked by real-life events.

Manuscript type: Empirical paper Research Aims: This paper aims to identify the topics of social media posts (on Twitter) from local skincare brands in Indonesia. Design/methodology/approach: This descriptive study applied text analysis methods, specifically frequency analysis and topic modeling with the Latent Dirichlet Allocation (LDA) method. Data were collected from the top four skincare brands with over 10,000 followers using Twitter Academic API interfaced through the Twarc package in Python. The dataset contained 1290 tweets from Wardah, 17,491 tweets from Avoskin, 1,968 tweets from Somethinc, and 590 tweets from Whitelab. The data analysis used quanteda and topicmodels packages from R software. Research findings: The findings revealed that the brands’ tweets can be categorised into 8-10 topics, which can be distilled further into three significant functions proposed by Saxton and Waters (2014): information, promotion and mobilisation, and community building and dialogue. Theoretical Contribution/Originality: This study adds to the body of knowledge regarding social media marketing by adding insights into how local Indonesian brands carry out their social media marketing efforts. Practitioner/Policy Implication: The findings implied that marketing managers should emphasise providing timely and relevant information as a base that can be used to build dialogic communication efforts with their audience. Research limitation/Implications: This study only looked at four brands in the local skincare product category. In addition, this study applied automated topic modeling with LDA, which needed an a priori number of topics, which might affect the topic labelling. Furthermore, this study’s scope was limited to categorising the brand posts without examining the effects of brand post types on engagement metrics. Future studies can examine more samples by observing other product categories, applying thematic analysis to code each tweet, and comparing the results to topic modeling results. Lastly, future studies can examine the relationship between brand post types/categories and engagement metrics such as likes, comments, and retweets. Keywords: Cosmetics Brands, Social Media Marketing, Skincare, Topic Modeling, Twitter

Manuscript type: Empirical paper Research Aims: This paper aims to identify the topics of social media posts (on Twitter) from local skincare brands in Indonesia. Design/methodology/approach: This descriptive study applied text analysis methods, specifically frequency analysis and topic modeling with the Latent Dirichlet Allocation (LDA) method. Data were collected from the top four skincare brands with over 10,000 followers using Twitter Academic API interfaced through the Twarc package in Python. The dataset contained 1290 tweets from Wardah, 17,491 tweets from Avoskin, 1,968 tweets from Somethinc, and 590 tweets from Whitelab. The data analysis used quanteda and topicmodels packages from R software. Research findings: The findings revealed that the brands’ tweets can be categorised into 8-10 topics, which can be distilled further into three significant functions proposed by Saxton and Waters (2014): information, promotion and mobilisation, and community building and dialogue. Theoretical Contribution/Originality: This study adds to the body of knowledge regarding social media marketing by adding insights into how local Indonesian brands carry out their social media marketing efforts. Practitioner/Policy Implication: The findings implied that marketing managers should emphasise providing timely and relevant information as a base that can be used to build dialogic communication efforts with their audience. Research limitation/Implications: This study only looked at four brands in the local skincare product category. In addition, this study applied automated topic modeling with LDA, which needed an a priori number of topics, which might affect the topic labelling. Furthermore, this study’s scope was limited to categorising the brand posts without examining the effects of brand post types on engagement metrics. Future studies can examine more samples by observing other product categories, applying thematic analysis to code each tweet, and comparing the results to topic modeling results. Lastly, future studies can examine the relationship between brand post types/categories and engagement metrics such as likes, comments, and retweets.

Abstract Multi-brand analysis based on review comments and ratings is a commonly used strategy to compare different brands in marketing. It can help consumers make more informed decisions and help marketers understand their brand’s position in the market. In this work, we propose a multifacet hierarchical sentiment-topic model (MH-STM) to detect brand-associated sentiment polarities towards multiple comparative aspects from online customer reviews. The proposed method is built on a unified generative framework that explains review words with a hierarchical brand-associated topic model and the overall polarity score with a regression model on the empirical topic distribution. Moreover, a novel hierarchical Pólya urn (HPU) scheme is proposed to enhance the topic-word association among topic hierarchy, such that the general topics shared by all brands are separated effectively from the unique topics specific to individual brands. The performance of the proposed method is evaluated on both synthetic data and two real-world review corpora. Experimental studies demonstrate that the proposed method can be effective in detecting reasonable topic hierarchy and deriving accurate brand-associated rankings on multi-aspects.

Public opinion expressed on social media increasingly mirrors consumer perception of brands in the Fast-Moving Consumer Goods (FMCG) sector. This study builds and compares sentiment analysis models to objectively classify public opinion toward three Danone Indonesia brands, namely AQUA, Mizone, and SGM, based on conversations on X/Twitter, while identifying the topics that shape that opinion. Data were collected legally through the official X/Twitter API v2, yielding 5,813 Indonesian-language tweets posted between May 2025 and June 2026 after duplicate removal. Three annotators labeled relevance and sentiment, with reliability assessed using Fleiss' Kappa, and the text was represented through TF-IDF weighting. Four classifiers, namely Naive Bayes, Support Vector Machine (SVM), Decision Tree, and Logistic Regression, were compared using cross validation, while Latent Dirichlet Allocation modeled the topics. After hyperparameter tuning and stratified ten-fold validation, SVM was the most accurate and stable model, reaching 88.80% accuracy and 88.19% macro F1. AQUA was dominated by negative opinion driven by a water-source controversy, whereas Mizone and SGM leaned positive. Integrating sentiment analysis and topic modeling objectively measures and explains public opinion toward FMCG brands.

This study examines fashion brands’ diversity, equity, and inclusion (DEI) progress by analyzing companies’ DEI statements using topic modeling and visual representations of fashion workers (models) using computer vision. The dimensionality of social inclusion (acceptance, belongingness, empowerment, equality, and respect) offers conceptual guidance to interpret DEI statements. Furthermore, using age as one of the diversity representations, we perform computer vision analyses to capture the age distribution of models in the photos of fashion brands. First, 121 DEI statements were scraped from brands’ official websites for topic modeling analysis. Three topics emerged from topic modeling – community, equality, and women-focused, which were consistent with the dimensionality of social inclusion. Second, 14,763 unique person images collected from WindowsWear were used for computer vision analyses. Results indicated that brands’ visual presentations lack the representation of advanced-age fashion workers in practice.