29 January 2018

In-Article Chatbots

By BBC News Labs

How can we help less engaged audiences understand big news stories?

The bigger a news story, the harder it gets for the casual reader to understand recent developments.

This is particularly a problem for readers that tend to be less engaged with the news. For example, our research shows younger readers may know that a story is important, but often don't know where to look for background information. This can make them feel nervous and anxious when they try to engage with current events.

The BBC's traditional explainer format of long Q&A-style articles is one way that journalists can provide background to less engaged audiences. However, these can stretch to over ten thousand words, with no way for readers to personalize the information they see on-screen. They also require readers to navigate away from a more recent news story to seek out the information they need.

We thought: wouldn't it be better if readers could simply ask for explanations on the elements of the story that they don't understand?

Our in-article bots are designed to feel like a conversation with a BBC expert. Users can choose from a list of questions to explore different elements of a story in more depth, based on their interests and current level of understanding of a topic. This prevents readers from feeling overwhelmed by large amounts of text - only some of which may be relevant to their interests.

Screenshot of flu epidemic bot

Our in-article bot answers questions about this winter's flu epidemic

The bots are embedded within BBC News stories, meaning that readers can find the information they need to understand the latest reporting directly on the article pages themselves.

In developing the new in-article format, we also wanted to make it easy for journalists and editors to create the bots without demanding big changes to their workflows. Our internal Bot Builder web application includes a feature for transforming long Q&A explainers into a bot conversation. This allows journalists to reformat dense, stand-alone pieces into embeddable components that can exist on multiple stories across the BBC News website.

Credit for our in-article bots goes to: Grant Heinrich, Tom Maslen, Alvin Ourrad, Kevin Peachey, Joy Roxas and Paul Sargeant.


Love data and code?

We're looking for talented developers to join our team.