Copy
The latest and greatest from SocialSensor

June 2012


 


Permanent link
Friend on Facebook Facebook
Follow on Twitter Twitter

WELCOME

Welcome to the first newsletter of the European project SocialSensor. Co‐funded by the EU 7th Framework Programme, SocialSensor is an Integrated Project that started its activities since the 1st of October 2011.

SocialSensor will develop a new framework for enabling real-time multimedia indexing and search in the Social Web. The project will move beyond conventional text-based indexing and retrieval models by mining and aggregating user inputs and content over multiple social networking sites. Social Indexing will incorporate information about the structure and activity of the users’ social network directly into the multimedia analysis and search process.

In the newsletter, which will be issued twice a year, we will present a synthesis of the principal news items and main activity of the SocialSensor project. In this context, we also like to point your attention to the SocialSensor website (http://www.socialsensor.eu/), which is regularly updated with news, announcements and research reports and results.

Finally, we’d like to invite you to register as a member of the SocialSensor User Group (SUG), which is a community of interest of relevant stakeholders from the Social Web. See the item below for a more detailed description of the SUG. To become a member please send a mail to contact@socialsensor.eu

TABLE OF CONTENTS:


Join the Social Sensor User Group (SUG)!

July 2012

The SUG is a community of relevant stakeholders from the Social Web including mobile application developers, researchers, news organizations, event organizers, social networks, advertisers and other relevant companies and users. SUG members will be informed about the latest SocialSensor progress and developments and after specific agreements it will be possible to participate in the evaluation of the project results, test the project software platforms and provide feedback. It will be possible to establish synergies for the possible exploitation of the project results, the development of business models, partnerships and participate in the technical discussions and activities (special sessions, standardization, etc) and contribute with ideas or requirements that may fit the project objectives. If you wish to become a member of the SUG, or have additional questions, please send an mail to contact@socialsensor.eu.

Like SocialSensor Newsletter 1 on Facebook Google Plus One Button share on Twitter

Social Web Industrial Workshop

July 2012

CERTH and Digital Spaces Living Lab organized on the 6th of July 2012 in CERTH, Thessaloniki, an industrial workshop on Social Media and Web Content Processing Services and Applications (Social Web 2012). The objective of the workshop was to gather industrial and academic stakeholders along with venture capitalists and encourage discussion and sharing of ideas in the areas of social media analysis and web content processing. A second aim of the workshop is to bring together venture capitalists, start-up companies and researchers to discuss emerging opportunities, obstacles and risks arising in entrepreneurship in these areas both at local and international level. There were 65 participants (42 from industry, 20 academics, 3 VCs), and the event was live broadcasted on the web.  Tweets about the event are available (for some time) at: 
https://twitter.com/#!/search/realtime/%23socialweb2012

Like SocialSensor Newsletter 1 on Facebook Google Plus One Button share on Twitter

Mobile photo browsing and tagging demo won the demo award @ ICMR 2012

June 2012

ClustTour, a city browsing and photo tagging mobile app, won the demo award at this year's International Conference on Multimedia Retrieval (ICMR 2012) in Hong Kong[1]. The app, which is available for free on the app store [2], supports efficient browsing and tagging of landmark and event photos. The application employs a hierarchical mode of exploration enabling zooming from the level of a city, through the level of an area/neighbourhood, down to the level of a specific spot. This navigation mechanism combined with photo clustering, and a dual map-list viewing mechanism, makes possible the browsing of hundreds of thousands of publicly available geotagged photos on a smartphone. To reduce clutter on the screen caused by the large number of markers on the map, an adaptive clustering strategy is employed. In addition to the advanced browsing capabilities, the application enables users to import their personal photos from Flickr and to easily annotate them by propagating to them metadata (location, tags) from the currently viewed object (area, landmark). The application currently supports more than 30 cities worldwide. The app development was supported by SocialSensor.
[1] http://www.icmr2012.org/awards.html
[2] http://itunes.apple.com/app/clusttour/id487608260?mt=8


Figure: Symeon Papadopoulos receives the demo award from Cees Snoek.
 
Like SocialSensor Newsletter 1 on Facebook Google Plus One Button share on Twitter

Workshop on Social Media Applications in News and Entertainment supported by SocialSensor

April 2012

 
SocialSensor supported the Workshop on Social Media Applications in News and Entertainment (SMANE2012), which was hosted by the WWW 2012 Conference in Lyon, France, April 16, 2012. Workshop proceedings are available here. The presentations slides are also available.



Like SocialSensor Newsletter 1 on Facebook Google Plus One Button share on Twitter

SocialSensor at WWW EU Projects session

April 2012

SocialSensor was presented at the EU Projects session of the World Wide Web Conference (WWW 2012), which was held in Lyon, France during 16-20 April. Please see the SocialSensor website for a more detailed description of the workshop and to download the presentation.

Like SocialSensor Newsletter 1 on Facebook Google Plus One Button share on Twitter

Business session in SMANE workshop: Social Media and News

April 2012

On the 16th of April 2012 (see link), Jochen Spangenberg (Deutsche Welle and SocialSensor project member) moderated a business session as part of the "International Workshop on Applications in Social Media News and Entertainment (SMANE 2012)", which was held in cooperation with the MSND workshop ("Mining Social Networks Dynamics. An introductory keynote presentation by Nick Newman, independent media consultant, former BBC employee and SocialSensor project member, portrayed how social media is affecting the business and practice of journalism. Please see the SocialSensor website for a more detailed description of the workshop. For further information see link.

Like SocialSensor Newsletter 1 on Facebook Google Plus One Button share on Twitter

Research Highlights


New approach for social event detection using multiple modalities

Giorgos Petkos, CERTH, Thessaloniki

One of the goals of SocialSensor is the development of methods that can automatically detect real-life events in content collected from the social media. CERTH recently presented a relevant paper at ICMR 2012 [1] (held in Hong Kong) entitled "Social Event Detection using Multimodal Clustering and Integrating Supervisory Signals" that proposes a new approach for detecting real world social events. It tackles the problem of social event detection as a problem of clustering social content items. This is a challenging problem, considering the fact that content from the social media may be expressed by heterogeneous modalities (e.g. text, multimedia, various types of metadata) that are not easy to process in a unified manner. The new approach utilizes an existing example clustering of items into groups that are related to social events, in order to guide the combination of the different modalities. The method essentially takes advantage of the example clustering to generate in an automatic manner a fusion strategy of the multiple modalities with which the content is associated. It has been applied on a set of photos collected from Flickr with excellent results, correctly grouping the items in sets which represent social events and outperforming a competing approach that utilizes a common fusion strategy. Moreover, the proposed approach is applicable to other multimodal clustering problems where a relevant example clustering may be available.
[1] http://www.icmr2012.org/


Figure: Example events detected by our method vs. baseline.

Like SocialSensor Newsletter 1 on Facebook Google Plus One Button share on Twitter

Experimental study of large-scale image search

Eleftherios Spyromitros-Xioufis, CERTH, Thessaloniki

Towards the development of a fast, large-scale image retrieval component, CERTH conducted an experimental study on efficient and accurate methods for retrieving images representing the same (or similar) object/scene with a given example image. The study, which was published at the 13th International Workshop on Image Analysis for Multimedia Interactive Services, WIAMIS 2012 [1], focused on an established pipeline for image representation and retrieval and evaluated the novel combination of SURF local image descriptors and VLAD vectors for descriptor aggregation. VLAD (Vector of Locally Aggregated Descriptors) is a recently proposed method for aggregating a set of local descriptors extracted from an image into a vectorized image representation. Its main advantage compared to other aggregation approaches such us Bag-of-Words (BOW) lies in the fact that it is able to provide very compact and highly descriptive image signatures which can be efficiently indexed, enabling accurate, real-time retrieval of near-duplicate images in collections containing hundreds of millions of images.  Originally, VLAD was proposed for the aggregation of SIFT features, however, SIFT extraction on an image example takes a significant amount of time and can be a bottleneck of the retrieval pipeline. Motivated by this limitation, the authors replaced SIFT with SURF features which are targeted for efficient extraction and are claimed to perform competitively to SIFT in a large domain of applications. The findings of the study were that the VLAD+SURF combination, not only reduces the total retrieval time but also exhibits better search accuracy than VLAD+SIFT in two publicly available benchmark image collections.
[1] http://wiamis.dcu.ie/


Figure: Example based image retrieval representing the same object/scene.

Like SocialSensor Newsletter 1 on Facebook Google Plus One Button share on Twitter

Explaining Trends on Twitter

David Corney, Carlos Martin, City University, London

One aim of SocialSensor is to identify trending topics in social media and put them in context. To explore this, we collected over 250,000 tweets sent on or around “Super Tuesday,” the day that ten USstates held elections to select the Republican Party presidential candidate. We calculated what fraction of tweets in each hour was associated with each candidate and with each state. As an example, this graph shows the distribution of tweets for one candidate, Newt Gingrich, in one state, Alaska.
Two peaks stand out, and we examined the words used most frequently in those tweets. The first peak contained many tweets discussing Gingrich’s recent speech about energy policy; the second peak followed a TV interview in which Sarah Palin (the former governor of Alaska) announced that she had just voted for Gingrich. By finding peaks in online activity and by highlighting key terms and tweets, we can use Twitter to identify important real-world events and place them in context, quickly and reliably.

Figure: Fraction of tweets per hour about Newt Gingrich in Alaska. The period covers Monday 5th March 15:00 -- Thursday 8th March 17:00 (GMT). The grey bar represents “Super Tuesday” (Alaskan time). Example tweets from two peaks in activity are also shown.

Like SocialSensor Newsletter 1 on Facebook Google Plus One Button share on Twitter

Upcoming Events


14 September 2012: Summer School on Social Media Modeling and Search


SocialSensor will organise in cooperation with CUBRIK the 2012 Summer School on Social Media Modeling and Search (SSMS 2012), on 10-14 September 2012 in Fira, Santorini, Greece. The summer school is aimed at PhD students, young researchers and practitioners interested in social media modeling and analysis, online social interaction analysis and relevant approaches and applications in prediction, search and visualization. Lectures include introductory and advanced courses by leading researchers from academia and industry, and will cover both theoretical and practical aspects of state-of-the-art research in these multi-disciplinary topics. SSMS 2012 is an ACM Chapter event of the EuroMM, the European Chapter of the ACM SIGMM.
Like SocialSensor Newsletter 1 on Facebook Google Plus One Button share on Twitter

25-26 September 2012: SocialSensor participates to the EC Search Computing Event


SocialSensor will participate with a presentation and demonstration to the International Search Computing workshop in Brussels organized by the CHORUS+ Network and the Media Search Cluster. The workshop will be a platform for the discussion of the key research, business and socio-economic issues in search computing. The workshop conclusions will contribute towards future research agendas and programmes, such as the European Union's Common Strategic Framework (Horizon 2020). The program is available here.
Like SocialSensor Newsletter 1 on Facebook Google Plus One Button share on Twitter


The research leading to these results has received funding from the European Community's Seventh Framework Programme (FP7/2007-2013) under grant agreement 287975
SHARE
Copyright © 2018 SocialSensor, All rights reserved.
Email Marketing Powered by Mailchimp
unsubscribe from this list | update subscription preferences