The field of science communication and topic modeling is witnessing significant developments, with a growing emphasis on alternative platforms and innovative methods for analyzing and interpreting large volumes of text data. Researchers are exploring new sources, such as social media platforms, to assess their viability and relevance for science communication, and are proposing novel evaluation measures and frameworks to improve the quality and accuracy of topic models. The use of machine learning techniques, such as sparse autoencoders and BERTopic models, is becoming increasingly prevalent, enabling the detection of deeper conceptual themes and more effective text generation. Furthermore, there is a growing recognition of the importance of addressing biases and limitations in existing datasets and models, such as publication imbalance and discontinuous coverage. Noteworthy papers in this area include:

• A study on Bluesky posts referencing scholarly articles, which highlights the platform's emerging role as a credible source for science communication and altmetrics.
• A paper introducing Mechanistic Topic Models, which enables controllable text generation using topic-based steering vectors and reveals deeper conceptual themes with expressive feature descriptions.