Special issue: When social media meets physical world

Abstract

GPS information to filter out the irrelevant images with respect to the query image. Recently, more and more research efforts have been dedicated to the aforementioned challenges and opportunities. Especially in the past three years, related papers are extensively published in ACM MM, SIGIR, WWW, and CVPR. Yet, a special issue focusing on this specific topic on related journals is still missing so far. Therefore, we see a timely opportunity to organize a special issue to bring together active researchers to share recent progress in this exciting area. Our goals are three-fold: (1) theories and applications on social multimedia computing related to location services; (2) survey on the progress of this area in the past years; (3) discuss emerging applications based on this newly scenarios. With a very selective review process, eventually in total 14 papers are selected into our special issue. One core problem in social media computing is the problem of image tagging. To handle the problem of image tagging in social media data, Nie et al. (Geo-location driven image tagging via cross-domain learning) introduced a geo-location driven image tagging scheme via cross-domain learning. The work by Li et al. (On the tag localization of web video) further extends the problem of geo-aware social media tagging from image domain to the video domain. Another work from Sun et al. (Social video annotation by combining features with a tri-adaptation approach) further proposed a novel social video annotation approach that combines multiple feature sets based on a tri-adaptation approach. In addition, the work by Liu et al. (Detecting and tagging users’ social circles in social media) presents a general algorithm for detecting users real friends in social media and dividing them into different social circles automatically according to the closeness of their relationships. Besides the linking across modalities (tagging and With the advent of Web 2.0 era, there is an explosive growth of social media data available on user-sharing websites such as Flickr, Youtube, and Zooomr. Besides the plain visual signals, more and more social media data are now associated with rich location context such as GPS tags, location identifications, or geographical and cultural cues, which benefit a wide variety of potential location-based multimedia applications such as landmark identification, location recognition, tourism recommendation, as well as photographing suggestion. Such unique combination of social and geographical tagging enables a more seamless perspective to dig out valuable insights from the user-sharing multimedia corpus in a large scale. Especially, coming with the ever growing popularization of smartphone cameras as well as instant message sharing services such as Twitter and Foursquare, there is a great potential for exploring cheaply available, large-scale social multimedia to assist and facilitate location related computing and services. Recent years have witness a great popularity of researches on the location-based multimedia modeling, learning, and application in social media analysis. Such location information makes the traditional difficult problems in multimedia content analysis become more tractable. For example, large-scale image repository can be significantly pruned if we use the