No abstract

Social media provides high-volume and real-time data, which has been broadly used in diverse applications in sales, marketing, disaster management, health surveillance, etc. However, distinguishing between noises and reliable information can be challenging, since social media, a user-generated content system, has a great number of users who update massive information every second. The rich information is not only included in the short textual content but also embedded in the images and videos. In this paper, we introduce an effective and efficient framework for event detection with social media data. The framework integrates both textual and imagery content in the hope to fully utilize the information. The approach has been demonstrated to be more accurate than the text-only approach by removing 58 (66.7%) false-positive events. The precision of event detection is improved by 6.5%. Besides, based on our analysis, we also look into the content of these images to further explore the space of social media studies. Finally, the closely related text and image from social media offer us a valuable text-image mapping, which can enable knowledge transfer between two media types.

No abstract

Social media has been broadly applied in many applications in sales, marketing, event detection, etc. With high-volume and real-time data, social media has also been used for disaster responses. However, distinguishing between rumors and reliable information can be challenging, since social media, a user-generated content system, has a great number of users who update massive information every second. Furthermore, the rich information is not only included in the short text content but also embedded in the images, videos. In this paper, to address the emerging challenge of disaster response, we introduce a reliable framework for disaster information understanding and response with a practice on Twitter. The framework integrates both textual and imagery content from tweets in hope to fully utilize the information. The text classifier is built to remove noises, which can achieve 0.92 F1-score in classifying individual tweet. The image classifier is constructed by fine-tuning pre-trained VGG-F network, which can achieve 90\% accuracy. The image classifier serves as a verifier in the pipeline to reject or confirm the detected events. The evaluation indicates that the verifier can significantly reduce false positive events. We also explore Twitter-based drought management system and infrastructure monitoring system to further study the impacts of imagery content on event detection systems and we are able to pinpoint additional benefits which can be gained from social media imagery content.

Social media provides high-volume and real-time data, which has been broadly used in diverse applications in sales, marketing, disaster management, health surveillance, etc. However, distinguishing between noises and reliable information can be challenging, since social media, a user-generated content system, has a great number of users who update massive information every second. The rich information is not only included in the short textual content but also embedded in the images and videos. In this paper, we introduce an effective and efficient framework for event detection with social media data. The framework integrates both textual and imagery content in the hope to fully utilize the information. The approach has been demonstrated to be more accurate than the text-only approach by removing 58 (66.7%) false-positive events. The precision of event detection is improved by 6.5%. Besides, based on our analysis, we also look into the content of these images to further explore the space of social media studies. Finally, the closely related text and image from social media offer us valuable text-image mapping, which can enable knowledge transfer between two media types.