Journeys in big data and AI across the transport networks of London & Paris

When looking for examples of digital innovation, few of us would think of public transport. But it turns out the sector is a rich source of use cases for big data, open APIs and even artificial intelligence. In a major city like London or Paris, there are millions of travelers every day who need accurate, timely information about the current schedules of thousands of buses, trams and trains. Getting that information to them on their smartphones has taken a lot of foresight, creativity and graft over the years, and the innovation continues today.

In one example launched in Britain last month, rail booking service Trainline has begun using AI to generate personalized alerts about travel disruption to users of its mobile app. There are quite a few moving parts to making all of this work. It starts with analysis of the Twitter feeds published by all of the UK’s various train operating companies. Trainline has figured out that news about disruption often gets on Twitter a lot faster than it takes to flow through the official national rail service data feeds. So if it can react to the Twitter feeds fast enough, it can get that information out to its customers faster.

AI, big data and voice tech in London

Using natural language processing, the AI automatically works out which tweets are important, and then analyzes where the disruption is having an effect. The next step is to match that analysis to individual journeys. This then feeds through to the Trainline voice app, which is designed for use by travelers on the move. They simply ask the app about their journey, and it will show them details of any disruption.

The combination of AI, big data and voice tech is typical of how providers are having to work with these cutting-edge technologies to serve the public transport user. The challenges include the large number of different data sources from multiple operators, the need to work with real-time data streams and map that data to a complex web of potential journey routes, and finally finding creative ways to deliver the information quickly and efficiently to travelers.

Analyzing tweets is quite a novel idea, whereas the concept of operators publishing data feeds about their services is now well established. Transport for London, the British capital’s transport authority, started publishing an open data feed of all its schedules in 2010, and it also provides live updates, for example showing the location of buses in transit. As in many cities around the world, these feeds inform the public transit journey planning on Google Maps and many other apps.

SNCF chatbots in Paris

In Paris, national rail operator SNCF is responsible for 6,200 trains a day that carry 3.2 million passengers on the Transilien network between the suburbs and the city center. It has been piloting the use of chatbots to give passengers tailor-made information about their journeys via Facebook Messenger, with the aim of rolling out a full service next year. Getting the information available in the right format has been a necessary first step, says Olivia Fischer, Digital Media Director:

A big part of the work for us at SNCF is to reorganize our IT system because, since it’s a very old company, the IT is also very old and it was built by layers to bring more information. We are really working hard at this time to reorganize and transform all this data and put it in an API.

In fact there are multiple APIs, she tells me – information about stations, service schedules and real-time operational data. It’s only this year that it’s become possible to blend real-time data with schedules, which allows the bot to give people updates about the status of specific trains.

It’s other media that forced us to organize [the data], but the chatbot arrives at a time where it’s almost the end of the story of this reorganization, and then we are able to deliver it through a chatbot.

Journey information and alerts

The current phase of the project has focused on training the bot so that it becomes familiar with how people ask it for information. It is built on technology by French startup Recast.ai, which was acquired early this year to become the basis of SAP Conversational AI. The next phase will offer full door-to-door journey information and alerts. Tourists and visitors will be a particular target market, since there’s no need to download a specific app or go to a website, and the chatbot will be able to operate in many different languages. This is where the technology adds value, says Fischer:

The chatbot reverses completely the relation because we go where the customers are and we don’t ask them to go to our media and to learn how to use our media. We go to them …

It’s the best way to have both a tailor made service and [provide it] on an industrial basis. Only the chatbot can do that.

My take

These different examples illustrate similar trends – a need to converge legacy data so that it can be consumed in a meaningful way via APIs, together with the use of AI services and tools to deliver timely, personalized information at scale. We often think of such projects being the domain of richly resourced commercial organizations in sectors such as financial services and high-end retail.

Public transport may not be so glamorous, but its operators and others in the sector have been early pioneers of real-time data streaming and analysis at scale to deliver invaluable individualized updates to large numbers of consumers on the move. Their achievements deserve recognition – and hold useful lessons for those in other industries following their lead.

Image credit – Young woman using mobile phone on public transport © Rawpixel.com – shutterstock Disclosure – SAP is a diginomica premier partner and funded the author’s travel to its Sapphire conference in Orlando where the interview with SNCF took place.