Journal of Visual Communication and Image Representation
Special Issue on Text-based Image/Video Understanding in Social Media Context
摘要截稿:
全文截稿: 2018-11-01
影响因子: 2.479
期刊难度:
CCF分类: C类
中科院JCR分区:
• 大类 : 计算机科学 - 3区
• 小类 : 计算机:信息系统 - 3区
• 小类 : 计算机:软件工程 - 3区
Overview
With the emergence and popularity of social networking services, such as Twitter, Facebook, and Tencent Weibo, a massive-scale user generated contents are posted every day through social platforms. During the past few years, we have witnessed an unprecedented growth in the volume of unstructured social media data. The social media big data has been naturally generated by the social networking services’ users. Nowadays, a rich number of images and videos are associated with textual information, such as user comments from Youtube and aesthetic annotations from Flickr photos. These textual cues are significant for interpreting the semantics of images/videos. For example, the state-of-the art weakly-supervised learning algorithms can effectively derive region-level semantic labels by leveraging image/video-level textural labels. Practically, accurately calculating image/video semantics by optimally discovering socially-aware texts is an indispensable technique in many intelligent systems, e.g., human action annotation/parsing, fine-grained visual recognition, and person re-identification.
Although generally conveyed in natural language, social media is distinguishable from the common textural labels in terms of style, tone, purpose, etc. For instance, items posted on Twitter usually include jargons, emoticons, or abbreviations which usually do not follow formal grammar, due to the length limitation. Hence, it is not suitable to apply the existing text processing techniques to model the content of social media data. Even worse, a few social media items are noisy or mistaken since users comments are quite objective. Thereby, accurately understanding image/video semantics by maximally alleviating the influence of noisy texts is a rigid demand. Furthermore, besides social media texts, there also exists multimodal information in user generated contents, such as images, audios, videos. In total, when socially-aware Internet-scale texts meets image/video understanding, many interesting issues and challenges will be created. We expect new mathematical formulations, datasets, and evaluation benchmarks to tackle image/video understanding problems in social media context.
This special issue serves as a forum to bring together active researchers from both industry and academia to exchange their opinions and experiences in image/video understanding under the framework of texts generated from social media. We solicit original contributions in threefold: (1) presenting state-of-the-art theories, techniques and novel applications of image/video understanding in social media; (2) surveying the recent progress in this area; and (3) establishing the benchmark corpora for corpus linguistics. This special issue target researchers and practitioners from both industry and academia.
The topics of interest include (but not limited to):
- Text normalization for social media, including misspellings, slang, profanity and neologisms;
- Image/video-based sentiment and emotion analysis in social media;
- Multi-view feature learning for text-based image/video modeling;
- Image/video analysis, retrieval, extraction, classification, tracking and summarization;
- Socially-aware image/video retrieval, classification and clustering;
- Learning to match and rank for social media;
- Social media recommendation for image/video applications;
- Socially-aware texts for video summarization;
- Image/video semantic calculation in noisy social media context;
- Social media textual entailment recognition, extraction and generation;
- One-round or multi-round conversation based social media;
- Image/video processing methods for domain-specific social media, such as healthcare, medical, aesthetics, etc.
- Multimodal socially-aware image/video corpus generation;
- Image/video annotation in social media by machine learning methods;
- The information fusion or relevance detection in multimodal social media;
- Harvesting image/video knowledge from social media;
- Knowledge graph based automatic annotation, question answering, summarization, etc.