Analysing social network structures and thematic engagement on X audio spaces

Abstract The rise of audio-based platforms such as X Spaces introduces synchronous, ephemeral modes of interaction that differ from traditional text-based social media. This study investigates user connectivity and thematic engagement by analysing a substantial dataset of X Spaces events. We model participation as a bipartite user-Space network, projected into user-user connections to assess community structure and interaction patterns. Our network analysis confirms broad-scale network characteristics and provides novel insights into local structures and user influence. Beyond structural analysis, we examine how users engage thematically across modalities by comparing creator-assigned topics, conversation-derived topics extracted from transcribed audio summaries, and textual posts. Our results reveal that user interests are only weakly aligned across modalities, indicating distinct communicative roles for audio and text. We also introduce a hybrid method combining BERT embeddings, spaCy similarity, and expert validation to assess the alignment between creator-assigned topics and actual conversation content. While most Spaces exhibit high topic coverage, 44% introduce additional themes, suggesting that live audio conversations often diverge from their predefined scope. These findings contribute to understanding interest expression, structural cohesion, and topical drift in emerging audio-based social platforms.