Comparing LDA/NMF/BERTopic on Twitter/X posts
The previous use case showed us the value of annotating social media posts. Since we have learned many other NLP techniques, can we apply these other techniques? In this use case, we will learn how they applied more techniques to the social media data.
Background
The richness of unstructured social media data has opened a new avenue for social science research. Topic modeling techniques have been applied to classify data and gain insights into it.
Questions
Similar to the previous use case on social media text, how can we apply NLP techniques to annotate a large number of social media posts?
NLP solution
The authors of [7] compared different types of topic modeling algorithms and documented their empirical findings. They collected 31,800 unique Twitter posts relating to travel and the COVID-19 pandemic. They applied Python LDA, NMF, and BERTopic modeling. They showed the standard results of these models, such as the topics...
