Extracting Media Items from Multiple Social Networks
2 likes1,991 views
This document discusses the extraction of media items from various social networks, focusing on the distinction between media items, microposts, and social networks. It highlights a media collector server implemented using Node.js that aggregates content from multiple platforms, along with an evaluation of its performance against other media galleries. The conclusion points out the potential for utilizing fresh media from social networks to tell event stories and outlines future work areas such as image de-duplication and entity recognition.
Extracting Media Items from Multiple Social Networks
- 1. What Fresh Media Are You Looking For? Extracting Media Items from Multiple Social Networks Giuseppe Rizzo1, Thomas Steiner2, Raphaël Troncy1, Ruben Verborgh3, José Luis Redondo Garcia1 and Rik Van de Walle3 <raphael.troncy@eurecom.fr> / @rtroncy 1 EURECOM, France 2 Google & University Politècnica de Catalunya, Spain 3 IBBT Ghent, Belgium
- 2. Conferences and natural disaster 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan -2
- 3. 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan -3
- 4. 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan -4
- 5. 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan -5
- 6. 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan -6
- 7. Some definitions Media Item: a photo or a video that is shared on a social network Micropost: a text status message that can optionally accompany a media item Social Network: an online service that focuses on building and reflecting social relationships among people sharing interests or activities Media Sharing Platforms: emphasis on sharing media but blurred boundaries with social networks since users are encouraged to react on media content (like, comment, favorite, etc.) 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan -7
- 8. Social networks and media items First-order support: Posting requires the inclusion of a media item Example: Flickr, YouTube Second-order support: Possibility to post media items but also text-only messages Example: Facebook Third-order support: No direct support for media items but rely on third party applications to host them Example: Twitter before the introduction of native photo support 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan -8
- 9. Media Collector (Server) Composition of media item extractors (12 SNs) Rely on search APIs + a fix 30s timeout window to provide results Fallback on screen scraping when necessary (Twitter ecosystem) Implemented as a NodeJS server Serialize results in a common schema (JSON) 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan -9
- 10. Deep link Permalink Clean text for NLP processing Aggregate view of ALL social interactions 12 Social Networks 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan - 10
- 11. Media Finder 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan - 11
- 12. Evaluation (1/3) 9 events occurring between 10 and 19 January 2012 Assad speech, CES Las Vegas, Costa Concordia Disaster, Cut the Rope Launch, Dixville Notch, Free mobile launch, Blackout SOPA, Ubuntu TV launch, Christian Wulff case 448 images + 143 videos Photo-Sweeper CBIR-based image duplication detection software Dataset heterogeneity: Leaderboard banner (728x90) to a standard 3.1 mega pixels (2048x1536) cell phone photo … no quadratic bitmaps shrinking Hard problem! Best settings for each event, no generic configuration, in order to limit the number of duplicate misses and false positives 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan - 12
- 13. Evaluation (2/3) User study to compare the relevance and illustrativeness of the media galleries One event: Google IO (“google i/o” + “io12”) http://en.wikipedia.org/wiki/Google_io Three systems: Media Finder, Twitter Gallery, Teleportd 7 participants (6 male, 1 female) in 2 groups MediaFinder Teleportd Twitter Google i/o 108 (49%) 20 (9%) 96 (44%) io12 69 (37%) 20 (10%) 98 (53%) 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan - 13
- 14. Evaluation (3/3) Q1: How illustrative this gallery is for this event? Q2: How visually diverse this gallery is for this event? Lickert 7-scale: result http://goo.gl/QzSM6 + http://goo.gl/7ov6Q google i/o io12 relevance Q1 Q2 relevance Q1 Q2 Media 0,28 2,35 2,72 0,21 2,05 2,24 Finder Teleportd 0,05 0,30 0,37 0,04 0,35 0,59 Twitter 0,28 2,64 2,64 0,34 3,44 2,91 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan - 14
- 15. Demo: Grid view 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan - 15
- 16. Demo: Timeline view 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan - 16
- 17. Conclusion Fresh media available on social networks Ignored by general search engines … … but ideal for building stories of events of our life Media Server: a NodeJS server collecting media items shared on social networks Media Finder: a client-server architecture that generates views of those media items http://mediafinder.eurecom.fr/ 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan - 17
- 18. Future Work Image de-duplication: Simple off-the-shelf tools using color, texture and shape (Ramaiah and Mohan, IEEE RAICS’11) Named Entity Recognition: NERD: http://nerd.eurecom.fr/ Clustering and Storyfying: Source and Temporal clustering Visual clustering Semantic clustering: using named entities extracted in microposts 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan - 18
- 19. http://www.slideshare.net/troncy 29/10/2012 - International Workshop on Socially-Aware Multimedia at ACM Multimedia 2012, Nara, Japan - 19