BBC News Labs kicked off MozFest week by holding a hackathon on conversational user interfaces with BBC Connected Studio.
Our #newsHACKS bring together some of the smartest people in the industry to build solutions to shared challenges. Many organisations have been experimenting with chatbots on platforms such as Facebook Messenger, WeChat and Line, but these tend to push headlines to users rather than provide new audience experiences. And with interest growing in voice devices like Google Home and Amazon Echo, it’s even more important for the media to develop new models of conversational journalism to engage audiences.
Thirteen teams of journalists, developers, UX designers and other industry professionals from across Europe joined us for our two-day hack to reimagine what interactions with the news could look like on these platforms and devices. Here’s a quick recap of what they built, what we learned and where we go from here.
We asked participants to submit prototypes in one of three categories:
- Editorial tools that help produce content for conversational interfaces;
- Unique audience experiences for conversational platforms or product;
- Surprise Us!
We were particularly interested in creative prototypes for voice devices.
A panel of four judges selected winners: Emily Withrow, editor of the Quartz Bot Studio; Leo Andrews, chief technical officer of Radioplayer; Varun Gujjanudu, product integration lead for Google Home in Europe; and Nathalie Malinarich, mobile and new formats editor for BBC News Online.
Best unique audience-facing experience: Sveriges Radio
Bringing the old-school elements of radio to an Echo device, the team from Sveriges Radio built a skill that allows users to save favourite songs and radio channels and share them with friends on social media platforms. Music lovers can ask Alexa to name the song that’s playing and add it to a Spotify playlist. News junkies can ask Alexa to play the news and favourite specific programmes. In both instances, users can ask Alexa to share content on Twitter, complete with a customised comment.
Best editorial tool: Voice Explainer BBC Ask
This BBC team built a tool for journalists that repurposes long Q&A articles for voice devices. The browser-based interface creates a question bank after being given a URL to a Q&A piece, stripping out the questions and their corresponding answers and displaying them as individual elements. Journalists can then edit the elements to better suit voice devices. The tool prompts users to edit answers down to a certain character limit and provides a text-to-speech preview feature to listen to the revised text being read aloud, as if by Alexa. The prototype also uses Google search queries to recommend related questions that can be answered by the replies stripped from the article.
Surprise us!: Realtime Audience Responses from the BBC Voice team
Reworking traditional audience polling and voting procedures, the BBC Voice Team developed an Alexa skill that captures voice commands in real-time and a visualisation tool that exposes the commands to editors. The tool records commands on a timeline that corresponds to a live programme’s run-time. Applications for the skill include voting on shows like Strictly Come Dancing and Eurovision and interacting with debates around local and national elections.
NRK — Noora: Integrating radio and text into an Alexa skill
This team from Norway built an Alexa skill that taps into both radio content and short text articles to provide users with news updates. Users can ask Alexa to play the news, whether there’s anything new happening today or request stories on a specific topic. When users request to hear the news, they are played a live radio feed; when requesting topical news, short text articles can also be served up via a text-to-speech service. The multilingual skill works in both English and Norwegian.
Bad Idea Factory: @TranscriptionBot
This duo-act built a Slack bot to aid journalists in editing, searching and discussing automatically transcribed interview content. Commands included:
- @bot transcribe: starts a transcription job on an uploaded audio file
- @bot play: assigns different Slack profiles to the different speakers in an interview and re-plays the conducted interview in the channel
- @bot summary: generates summaries of interviewee answers to the interviewer’s questions
BBC Broadcasting Systems Development: Fukushima Bots — Conversational Text Adventures
This prototype takes situational and educational articles and repurposes them as interactive narratives. Showcasing the idea with an article on Japan’s Fukushima reactors, the team demoed how users can explore a story’s environment and make decisions in a non-linear narrative. Akin to gameified journalism, the prototype could also be developed to display multimedia content on a second screen device, according to the team.
BBC Me: All Day, Every Day
This team synchronised an Alexa skill with a BBC Me application to serve up recommended content to users throughout the day. A user begins the morning by asking Alexa for the top stories, which they can then navigate through by skipping or saving for later, depending on their interest. Saved stories are pushed both to the BBC Me application and in an email for offline reading during the user’s morning commute. Based on the stories the user saved, the BBC Me app creates a playlist of suggested content on similar topics, which can be played before bed and paused by signals from a wearable fitness tracker when the user falls asleep.
Financial Times: FTConversation
What’s more conversational than reader comments? The team from the Financial Times built an action for Google Home called FTConversation that allows readers to submit their thoughts on articles they’ve read and listen to comments recorded by other readers. To implement a degree of quality control, users must answer a question showing they’ve read and understood an article before recording a comment. Sentiment analysis and swear word detection work to automate the moderation process — if you’re caught cursing, you’re charged for your comment or can take the opportunity to clean it up and re-record.
Telegraph and Free Press Unlimited: Engaging with audiences using Facebook Messenger
This tool for enhancing audience engagement allows news organisations’ audiences to send stories to journalists via a Facebook Messenger bot. A custom-built dashboard displays information reported to the bot — including word clusters based on entity extraction and sentiment analysis on audience mood. The dashboard also categorises stories submitted by users, giving editors and journalists an easy view of what topics are popular among their readers.
Deutsche Welle: Stammtisch
The team from Deutsche Welle demoed a prototype bringing Stammertisch — the German word for gatherings of people discussing politics in pubs—to a voice assistant. The conversation kicks off after a user gets an overview of trending news and stories. Alexa then asks for the listener’s opinion on the news of the day, which they can record by speaking into their device (this was entered textually in the team’s demo). The user can request the recorded opinions of others in their peer network who have listened to the same story, and choose to engage with those whose opinions differ.
News Labs & friends: Knowing Me Knowing You
Demoing a new skill called BBC Ask an Expert, this combination BBC Voice / BBC News Labs team chose to put users in direct contact with interviewees to answer their questions. Bonus features include the ability to push the original media content directly to an app on the user’s phone.
Near East Digital
This combination BBC News Labs/Near East Digital team has asked that we not share their prototype.
BBC News Labs: To Me To You
This Alexa skill for Echo Show devices eliminates the need for page-long “Movies to Watch” lists by allowing friends to suggest shows to one another and storing those suggestions on iPlayer. Users can record a custom video message recommending BBC content to their friends, which their friends can subsequently play after work before diving directly into the recommended programme.
Editor of Labs Robert McKenzie and I will be facilitating a session on how to transfer learnings from a hackathon back into the office at the OpenNews Unconference on Journalism Technology on Thursday. We’re hoping the discussion will lead to a handful of agreed-upon practices to help newsrooms make the most of the ideas, code and skills that staff develop by taking part in these events.
Want to share ideas? Send us an email at email@example.com.
Our next #newsHACK is scheduled for November 22 and 23 in London. We’re hosting this event in collaboration with SUMMA, a pan-European project developing a new multilingual media monitoring and summarisation platform for newsrooms. News doesn’t break in just one language, and we’re interested in how participants can leverage SUMMA’s API to expose patterns to editorial teams. You can find more information and register on Eventbrite.