REVEAL Workshop Review & Resources
On 20 Sep 2016 the REVEAL project, in which DW Innovation is heavily involved, conducted its final workshop. The full-day event was organised by the project consortium under the leadership of ATC’s iLab and DW Innovation Projects. Topic of the day: social newsgathering and verification of eyewitness media.
Almost 90 participants attended and joined in lively discussions about challenges and approaches regarding the use and verification of social media in news. Judging by the feedback received, and following comments on Twitter under the conference hashtag #smvw16, the day turned out to be a big success!
Two keynotes speeches kick off the workshop
Sam primarily talked about how social media has changed the business of newsgathering. He stressed in particular that more care should be taken of the “uploader” – in other words the person providing content to social networks.
One of journalists’ prime tasks and concerns should be to protect and safeguard the uploader. Otherwise, there is the risk of serious damage to the individual (trauma, stress and such like) and – if not treated with respect and care – there is the danger of people stopping to share content that can be of great value to the newsgathering process and situation assessment.
Dhruti Shah of the BBC then focussed her keynote, the second of the day, on journalistic practices and the challenges journalists are confronted with if they are assigned to deal with content posted in social networks. In particular, Dhruti pointed to what she referred to as “journalism 101″: the fact that traditional, established journalistic principles still apply, namely to find answers to the “big journalistic questions” who, what, when, where, how and why.
Dhruti had furthermore brought along “a bag full of Internet”. It was filled with a variety of items, symbolising the treasures that can be found on the net. However, when Dhruti pulled out a doll that had its head ripped off, the laughter soon subsided: “Parts of the Internet are broken”, Dhruti concluded.
Having set the scene with these two excellent keynotes, next came presentations of the EC co-funded projects REVEAL, PHEME and InVID. Each project presented some of its highlights, demoed what has been achieved to date, and portrayed which challenges are tackled with respective solutions and approaches. An outlook on work ahead and what to expect as ultimate project results was also given in the sesions that followed.
REVEAL presentation and demos
The REVEAL overview was presented by Nikos Sarris of the ATC iLab. Nikos demoed the integrated REVEAL news application (not yet available for public / unguided testing) and showed the capabilities and functionalities of Truthnest, a stand-alone Tweet analysis and verification tool that is also integrated with the REVEAL application (if you want to trial and test Truthnest, get in touch with Nikos Sarris and send him your Twitter handle).
Next came Symeon Papadopoulos of CERTH-ITI who presented the work done in the area of image analysis and image verification. Results were demoed live. For anybody who is interested: you can test the tools yourself. These are:
- The “Disturbing Image Detector” that detects potentially traumatic images;
- The “Image Verification Assistant” that analyses images using a variety of algorithms. Results are provided individually;
- The “Tweet Verification Assistant” that checks tweets that contain images or video and determines their veracity;
- The “Popularity Prediction Demo”: a set of methods that analyse and predict the future values of popularity indices of news posts using a variety of features (the sources are all freely available on GitHub).
Symeon was followed by Anastasia Krithara who presented the work of NCSR Demokritos in the area of semantic search and relations extraction. Anastasia, too, had brought along a demo. If you are interested in taking a closer look, go to the Semantic Search demo and test it yourself.
Aleksandra Kuczerawy of CiTiP at the University of Leuven then rounded off the REVEAL session with some legal advice and recommendations, stressing that not everything that is technically possible is also legally allowed. In particular, and according to Aleksandra, more awareness is needed in the areas of data privacy, data processing and data protection. (The best overview of work done in the legal domain can be found in the respective reports – they are accessible via the Deliverables page on the REVEAL project website).
PHEME presentation and demos
PHEME is a project that combines big data analytics with advanced linguistic and visual methods. At the workshop, it was introduced with a presentation by the project coordinator Kalina Bontcheva of the University of Sheffield / Gate UK.
Then, Georgi Georgiev of Ontotext presented Linked Open Data-based concept extraction, rumour detection, and publishing applications. Here’s a semantic enrichment of a news demo developed by Ontotext which you can test yourself.
Arno Scharl of webLyzard came next. In addition to his presentation, Arno gave a demo of the PHEME veracity visualisation dashboard that is available for trialling here.
As in the previous session, presentations were followed by a round of Questions & Answers that, again, showed a huge interest in the topic generally, and the presented solutions in particular.
InVID presentation and demos
The final project presentation came from the InVID project – an EU co-funded project that deals in particular with the verification of videos.
InVID was introduced by Vasileios Mezaris of CERTH-ITI with a presentation outlining aims and approach of the project.
Then came Arno Scharl, as webLyzard is involved in both PHEME and InVID. Arno presented the system architecture of the multimodal analytics dashboard as it is planned to be developed in InVID.
Denis Teyssou of French news agency AFP subsequently showcased work done to date in determining the authenticity of videos posted on the web. His hands-on demo (screencast with commentary) is publicly available and can be viewed on YouTube, while the presentation is available here.
Rounding off with a round table session (that did not include a table…;)
The final session was a “round table” (without a table – so it was rather a “semi-circle” discussion) in which the keynote speakers Dhruti Shah and Sam Dubberley participated, completed by two representatives each of the research projects PHEME, InVID and REVEAL. Like the rest of the day, this session, too, was moderated by the author of this post (Jochen Spangenberg).
Some takeaways (small selection):
- Journalists are impatient creatures. “They want it fast!” And easy. Almost more importantly: journalists want it accurately! The more support they can get in verifying social media content, the better. However, the ultimate responsibility for deciding whether to publish something or not is likely to remain with humans, at least in the foreseeable future. (“Keeping the human in the loop.”)
- Being aware of potentially disturbing material and the impact this may have on both journalists as well as users is vital. Journalists should be provided with support and places / people to turn to in case of it affecting their well-being. More needs to be done here by media organisations to prevent trauma and negative effects on journalists. (A useful resource in this context is the work of the DART Center for Journalism & Trauma at Columbia Journalism School and Eyewitness Media Hub).
- Establishing codes of practice or conduct when it comes to contributors of social media content were regarded as very important, too. While first steps have been taken in this direction (e.g. the ONA Social Newsgathering Ethics Code, work of First Draft News in this area, and the research carried out by Eyewitness Media Hub, to name but a few), more is still necessary.
- Technology can greatly assist in social newsgathering and verifying social media. There are numerous tools out there, many of them free of charge, that aid tremendously in gathering and checking information that is posted by eyewitnesses of events. However, most of these tools are not specifically “made for journalists” and their needs. Furthermore, there are different tools for different tasks. There is big potential for tools and services that provide certain functionalities and take into account specifics of journalistic needs and requirements.
- It was stressed at various points that the solutions as they were presented throughout the day have great potential and are in high demand by journalists. “When can we start using what you are developing in our editorial work?”, was a question that was voiced frequently.
- Besides all the technology that could be of help, other aspects and practices (many of them “old-fashioned” – but nevertheless highly relevant and useful) matter, too, and must not be neglected. So do not forget or discard useful and worthwhile journalistic principles.
- Having good networks of people to turn to and ask for support or assistance is another vital resource. Dhruti Shah brought this to the point: “everyone needs friends” – the more the better! Verification often becomes easier if done collaboratively, with individuals bringing in specific skill sets (such as language skills, know-how of local areas and such like).
- When it came to panel participants voicing their wishes in the final round of comments, one particular remark featured repeatedly and got much support by the audience: having a high quality reverse video search tool (something TinEyeand Google Reverse Image Search provide for still images) would be extremely beneficial and highly valued. This also indicated that InVID is on the right track – and there are high hopes regarding its outcomes…
Trying to sum up and describe in a few key words or phrases what matters most in the field of social newsgathering and reporting, the panel participants named the following:
- efficiency (in dealing with information);
- accuracy (being as accurate as possible, and having “the right” technology to support this);
- handling and filtering large amounts of data and information (providing what is required for specific tasks; presenting this in an “easy to digest” way, e.g. by intuitive visualisations);
- responsibility (dealing with contributors fairly and taking care of the audience, while also guaranteeing the well-being of journalists).
After a long and intense day the workshop closed at 5.45 pm. Tool demos and discussions continued for a bit longer. The day was rounded off with a networking dinner in the venue’s garden. All in all: a successful day with happy but exhausted organisers, and a grateful audience.
Thanks to everyone who participated and made all this possible! For more impressions of the workshop please visit the REVEAL website.
Notes and disclaimers:
- The services that are linked in this post and are available for trialling are still “under construction”. There may be down-times at various stages, or components may not be fully functional. If this is the case, please try again at a later time / date. No responsibility of whatever nature is taken for malfunction, system impact and such like.
- If you want to get in touch with any of the people mentioned in this article, you are welcome to do so. Either contact them via direct message on Twitter, or get in touch with the DW team (Ruben Bouwmeester or Jochen Spangenberg) who are happy to forward your request to the respective person.
- Some further material will probably be added to this post as it becomes available (e.g. links, presentations, demos etc).
- This article was first published on the REVEAL project website.
DateFriday, September 23rd, 2016