Automatic Content Analysis of Social Media Short Texts: Scoping Review of Methods and Tools

Kasperiūnienė, Judita; Briedienė, Monika; Žydžiūnaitė, Vilma

doi:10.1007/978-3-030-31787-4_7

Cited by 7 publications

(6 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Despite the productive and unexpressed nature of compound words they often complicate computational analysis (Krishna et al, 2016). Other difficulties emerge when data are meaningless (i.e., noisy data) or when there are many gaps present in the data (i.e., sparse data; Kasperiuniene et al, 2020).…”

Section: Topic Modeling As a Solution To Cope With Unstructured Text ...mentioning

confidence: 99%

“…In the case of this study, Twitter was selected due to its strict regulations on the number of characters allowed per tweet, making it an ideal platform for exploratory research. Nonetheless, the methodological approach in this study should be applicable to other channels as well since social media posts, in general, are short and unstructured (Kasperiuniene et al, 2020). However, it is still critical to note that the nature of social media differs in terms of user demographics, text presentation, or rhetoric, amongst others.…”

Section: Limitations and Recommendations For Future Researchmentioning

confidence: 99%

See 1 more Smart Citation

A Topic Modeling Comparison Between LDA, NMF, Top2Vec, and BERTopic to Demystify Twitter Posts

Egger

2022

Front. Sociol.

353

View full text Add to dashboard Cite

The richness of social media data has opened a new avenue for social science research to gain insights into human behaviors and experiences. In particular, emerging data-driven approaches relying on topic models provide entirely new perspectives on interpreting social phenomena. However, the short, text-heavy, and unstructured nature of social media content often leads to methodological challenges in both data collection and analysis. In order to bridge the developing field of computational science and empirical social research, this study aims to evaluate the performance of four topic modeling techniques; namely latent Dirichlet allocation (LDA), non-negative matrix factorization (NMF), Top2Vec, and BERTopic. In view of the interplay between human relations and digital media, this research takes Twitter posts as the reference point and assesses the performance of different algorithms concerning their strengths and weaknesses in a social science context. Based on certain details during the analytical procedures and on quality issues, this research sheds light on the efficacy of using BERTopic and NMF to analyze Twitter data.

show abstract

Section: Topic Modeling As a Solution To Cope With Unstructured Text ...mentioning

confidence: 99%

Section: Limitations and Recommendations For Future Researchmentioning

confidence: 99%

A Topic Modeling Comparison Between LDA, NMF, Top2Vec, and BERTopic to Demystify Twitter Posts

Egger

2022

Front. Sociol.

353

View full text Add to dashboard Cite

show abstract

“…Although not without its critics (Bruinsma and Gemenis, 2019), Wordscores remains a popular computational text analysis method with multiple applications in research, being used to extract scores for new text parts or segments on the basis of scores for words derived from documents with known scores (Kasperiuniene et al, 2019). As such, it is an example of a supervised text-scaling algorithm, in that it requires certain texts to be assigned position scores by human annotators (Nanni et al, 2022).…”

Section: Discussionmentioning

confidence: 99%

Quantitative Textual Analysis as a means to explore corporate interests in food safety

Reeves

2023

Health (London)

View full text Add to dashboard Cite

The growing body of scholarship on the commercial determinants of health has, so far, mostly employed qualitative methods but this is now being complemented by a small, yet growing, corpus of quantitative studies. We illustrate the use of one such method, quantitative text analysis (QTA), in a case study of submissions to a public consultation on a draft scientific opinion by the European Food Safety Authority on the chemical acrylamide, demonstrating how this method can be used and insights that might be drawn from it. We use Wordscores as one example of QTA to illuminate the diverse positions taken by actors submitting comments and then assess whether the final policy documents moved towards or away from the positions taken by different stakeholders. We find a broadly uniform position among the public health community, opposed to acrylamide, contrasting with industry positions that were not monolithic. Some firms recommended major amendments to the guidance, largely reflecting the impact on their practices, while policy innovators seeking ways to reduce acrylamide in foods aligned with the public health community. We also find no clear movement in the policy guidance, likely because most submissions supported the draft document. Many governments are required to conduct public consultations, some attracting enormous numbers of responses, with little guidance on how best to synthesise the responses so the default position is often a count of those for and against. We argue that QTA, primarily a research tool, might usefully be applied in analysing public consultation responses to understand better the positions taken by different actors.

show abstract

“…Reliance on machine learning, however, can lead to misclassification (Lyutov et al 2021), so we examine the results of the topic modeling to identify from where the newly emergent topic stems and describe them. Many recent studies that apply BERTopic have performed qualitative or manual validation of the results (Balcı et al 2023;Capra, 2024;de Lima et al 2023;Kasperiuniene et al 2020;. Using qualitative analysis, we review the results of the BERTopic process to validate them.…”

Section: Embedded Topic Modelingmentioning

confidence: 99%

“…Qualitative validation of results. Following recent studies that apply BERTopic (Balcı et al 2023;Capra 2024;de Lima et al 2023;Kasperiuniene et al 2020;, this study performed qualitative or manual validation of the results. While topic modeling may allow for the analysis of a large corpus of data, the results of the topic modeling should remain decipherable to non-experts ).…”

Section: Articlementioning

confidence: 99%

Identifying interdisciplinary emergence in the science of science: combination of network analysis and BERTopic

Kim,

Kogler,

Maliphol

2024

Humanit Soc Sci Commun

View full text Add to dashboard Cite

Global scientific output is expanding exponentially, which in turn calls for a better understanding of the science of science and especially how the boundaries of scientific fields expand through processes of emergence. The present study proposes the application of embedded topic modeling techniques to identify new emerging science via knowledge recombination activities as evidenced through the analysis of research publication metadata. First, a dataset is constructed from metadata derived from the Web of Science Core Collection database. The dataset is then used to generate a global map representing a categorical scientific co-occurrence network. A research field is defined as interdisciplinary when multiple science categories are listed in its description. Second, the co-occurrence networks are subsequently compared between periods to determine changing patterns of influence in light of interdisciplinarity. Third, embedded topic modeling enables unsupervised association of interdisciplinary classification. We present the results of the analysis to demonstrate the emergence of global interdisciplinary sciences and further we perform qualitative validation on the results to identify what the sources of the emergent areas are. Based on these results, we discuss potential applications for identifying emergence through the merging of global interdisciplinary domains.

show abstract

Automatic Content Analysis of Social Media Short Texts: Scoping Review of Methods and Tools

Cited by 7 publications

References 44 publications

A Topic Modeling Comparison Between LDA, NMF, Top2Vec, and BERTopic to Demystify Twitter Posts

A Topic Modeling Comparison Between LDA, NMF, Top2Vec, and BERTopic to Demystify Twitter Posts

Quantitative Textual Analysis as a means to explore corporate interests in food safety

Identifying interdisciplinary emergence in the science of science: combination of network analysis and BERTopic

Contact Info

Product

Resources

About