In the ever-evolving field of social media analysis, determining the origins of tweets has become increasingly challenging due to contextual nuances. This paper addresses the fundamental problem of inferring spatial and temporal properties of social media posts, ensuring that these inferences accurately represent the actual event’s location and time. We introduce key concepts, such as ‘temporally inferred true locations,’ to distinguish between temporal and spatial inferences. Our analysis reveals significant findings: 9.95% of geotagged tweets are ‘Spatially Verified,’ while only 0.53% are ‘Temporally True Inferred Location.’ Additionally, 28.95% of posts fall under ‘Spatially Valid, Time Inferred,’ highlighting the affinity of time and space in social media narratives. Investigating non-geotagged tweets, we identify that 12.3% of these posts can be classified as ‘Spatially and Temporally Valid,’ demonstrating the potential of advanced location and time inference techniques, even without explicit geotags. In summary, our study not only unravels the intricacies of geotagged tweets but also emphasizes their crucial role in advancing location inference from digital narratives, addressing the challenges in inferring the true whereabouts and valid time of events described in social media posts.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Unveiling GeoX Posts: Advancing Spatial and Temporal Inference from Social Media Narratives

  • Iman Sukaiti,
  • Imad Afyouni,
  • Zaher Al Aghbari

摘要

In the ever-evolving field of social media analysis, determining the origins of tweets has become increasingly challenging due to contextual nuances. This paper addresses the fundamental problem of inferring spatial and temporal properties of social media posts, ensuring that these inferences accurately represent the actual event’s location and time. We introduce key concepts, such as ‘temporally inferred true locations,’ to distinguish between temporal and spatial inferences. Our analysis reveals significant findings: 9.95% of geotagged tweets are ‘Spatially Verified,’ while only 0.53% are ‘Temporally True Inferred Location.’ Additionally, 28.95% of posts fall under ‘Spatially Valid, Time Inferred,’ highlighting the affinity of time and space in social media narratives. Investigating non-geotagged tweets, we identify that 12.3% of these posts can be classified as ‘Spatially and Temporally Valid,’ demonstrating the potential of advanced location and time inference techniques, even without explicit geotags. In summary, our study not only unravels the intricacies of geotagged tweets but also emphasizes their crucial role in advancing location inference from digital narratives, addressing the challenges in inferring the true whereabouts and valid time of events described in social media posts.