The first hackathon from Al Jazeera's burgeoning media innovation community, Canvas, was held over the weekend, with applicants from around the world experimenting around the concept of 'media in context'.
Sixteen hundred people applied of which just 86 – including developers from MIT, the head of Hacks/Hackers South Africa, and the Pulitzer Prize-winning founder of Politifact – were invited to Doha to take part
Powered by 250 cans of Red Bull and barrels of coffee, here are some of our favourite ideas from the finished projects.
Screenshot from Street Stories
When Yana Kunichoff was reporting from Ferguson, Missouri this past summer, she felt her coverage was lacking the true sense of tension and urgency felt on the ground during the protests.
Inspired by her time there, the team created Street Stories as an "immersive narrative experience".
Users can explore the scene of a story using Google Streetview with important locations annotated with pictures, video, tweets or text by the journalist.
For ongoing or time-dependent stories, users can click through the days of the events to further explore what happened.
Street Stories won the award for "best consumption phase solution".
Screenshot from Narrata
The team behind Narrata – a sharp portmanteau of narrate and data – recognised how important visual storytelling has become, and made a web app to display interactive graphs or maps that move alongside text.
So far, so what? Where this differs from animated graphs and maps on show elsewhere is the Narrata team have taken the process out of the hands of developers and delivered it to the less technically proficient, so all a journalist needs is their data and their story and the program will do the rest.
The graphs and maps move as the reader scrolls and they can investigate the data independently. The upload tool is as simple and effective as Datawrapper.
Narrata won the award for "best editorial context solution".
Video demo of Newsclip.se (no sound)
Newsclip.se was presented as a "contextual research editor" that gathers relevant information related to keywords in an article from Wikipedia, the wider web and a growing, internal database.
What that actually means in a practical sense, is that journalists have a plugin for their CMS so the next time you're writing about a similar project, "all that information can come to you", said developer Bruno Faviero.
In the wider context of a news organisation, this strengthens an outlet's "institutional memory", he said, so all reporters have access to all the information and if someone is picking up a new beat they can get up to date quickly.
Newsclip.se won the award for "greatest potential for use".
Screenshots of the app, shared by the Reporta team
Built for Google Glass as well as the web, Reporta was designed to be a "personal reporter" to make news distribution a more interactive, personalised experience for what readers really want.
In the demonstration, a Glass-wearing team member had a live – if stilted – conversation with the Reporta app as it read article headlines and offered to save the full story to the user's Kindle.
Natural language processing, grammar patterns, voice recognition and artificial intelligence all play a part in how Reporta delivers the headlines in a conversational manner and remembers what types of stories the user prefers.
Reporta won two awards, for "highest design and technical quality" and "best personal context solution".
Screenshot from MapCake
MapCake intends to give greater "geographic context" to news stories to provide a better understanding of the issue, and also suggest geographically-related content.
Using a text editor and a map, journalists can tag keywords or phrases from their article to mark them as locations on a map and show related stories using Google geolocation software.
In their working example, they used the New York Times API to map topics and locations from a story about Ferguson, Missouri but in the future they said they could automate the process, and potentially integrate social media feeds with geolocated stories for publications.
MapCake won the award for "best situational context solution".
Image from Lasertag
The Lasertag team sought to "reanimate your media archive" by making it easier to navigate and search archived stories, pictures or videos, and link them to new articles being written.
The software looks for noun phrases and key words to automatically categorise an article, and the team developed a WordPress plug-in to act like a spellchecker during the writing process: highlighting words and suggesting related archive content.
Lasertag won the award for "best production phase solution".
A video demo of Near
Near was pitched in the "media on the go" challenge for people who might prefer listening to news over reading it, or for busy commuters who want to hear the news as they travel.
Short for "news delivered to your ear", Near reads headlines from a particular news organisation and responds to your commands: "next" to skip it, "bookmark" to save it, or "read more" to give you the full story.
In the future, the Near team want to allow curated playlists of news or voice-based searches relevant to the user's location.
Unlike print or the SEO-focus of the previous decade, headline writing for digital news in 2014 means optimising for social media.
The team behind Headitor recognised how time-consuming headline writing can be for some outlets, in testing different headlines for different audiences, platforms and times of the day.
They envisioned a program to provide "dynamic, personalised headlines for a publication's websites and apps".
Taking 10,000 stories from Al Jazeera and the New York Times, they cross-referenced the stories with social media to see which headlines performed better among which demographics.
At the time of presentation, Headitor could apparently write headlines for specific social networks and age groups but the team wanted to build on the programming so the algorithm will improve itself in writing better headlines.
A video demo of the Perspectiv.es platform (this video has no sound)
One story on a subject will never be the full story, and different outlets regularly have different views on the same event.
Based on this concept, the team behind Perspectiv.es built a Chrome extension that looks for "keywords and noun phrases" in an article to identify articles on the same subject at other outlets.
These are then presented to the reader, prioritised by those that are most different and in the hope that the tool will "legitimise other perspectives" and "humanise other people and viewpoints we might not be aware of".
Perspectiv.es won the "most innovative" award.
Safety in numbers
Starting a presentation in front of judges and potential employers with "YOU SUCK" was brave of the team behind Safety in Numbers, but neatly exemplified the issue they want to solve: negative comments.
The proposed solution is to turn over comment moderation to a network of users across a number of different sites.
The actions of one moderator would benefit the network as a whole and the software, already a WordPress plug-in, "weighs" a commenter's actions depending on whether they are a known and trusted member of the community, a bot or a regular offender.
More information on the challenges and criteria for the hackathon can be found on the Canvas website, along with all the presentations from the 19 teams. The full list of winners is as follows:
Most innovative: Perspectiv.es
Greatest potential for use: Newsclip.se
Greatest impact and disruption: Tiny FM
Highest technical and design quality: Reporta
Best embodiment of values and ethics: Acumen
Best production phase solution: Lasertag
Best distribution phase solution: Tiny FM
Best consumption phase solution: Street Stories
Best personal context solution: Reporta
Best situational context solution: Hakawati
Best editorial context solution: Narrata