Differing Content and Language Based on Poster-Patient Relationships on the Chinese Social Media Platform Weibo: Text...

Area of Science:

• Computational linguistics
• Social media analysis
• Oncology communication

Background:

• Breast cancer significantly impacts patients and their social circles.
• Social media platforms like Weibo are used by affected individuals to share experiences.
• Narratives on social media vary based on the poster's relationship to the patient.

Purpose of the Study:

• To explore differences in content and language on Weibo based on poster-patient relationships.
• To analyze variations in emotional expression and topics discussed.
• To differentiate posts made by patients versus those by friends, family, or acquaintances.

Main Methods:

• Utilized Weibo data, collecting 10,322 breast cancer-related posts.
• Employed a two-step Chinese BERT (Bidirectional Encoder Representations from Transformers) model for classifying poster-patient relationships.
• Conducted sentiment analysis using the Linguistic Inquiry and Word Count lexicon and topic modeling with BERTopic.

Main Results:

• Achieved high accuracy in classifying poster-patient relationships (F1-scores of 0.92 for binary and 0.83 for multiclass models).
• Identified significant differences in emotional expression: higher negative emotions and anger in 'no_patient' posts, and higher sadness and anxiety in 'family_members' posts.
• Found distinct topic distributions: 'no_patient' posts focused more on fears and anger, while 'family_members' posts emphasized cancer treatments.

Conclusions:

• Content and emotional expression in breast cancer posts on Weibo differ significantly based on the poster-patient relationship.
• Posts from family members show more sadness, anxiety, and focus on treatment, potentially aiding emotion regulation and support.
• Posts where the patient is not directly identified exhibit higher levels of anger, possibly serving as rants for emotional release.