Booking a hotel can be stressful, let alone booking an entire holiday. From pricing to location, facilities and all the extras you might need how do you know where to start and how to narrow it down? Websites such as Booking.com, the largest travel website in the world, have needed to find a way to create the ease and personalisation that would be found in a travel agent from the comfort of the user's own home. Booking.com have a team of more than 100 data scientists working on things such as product development, customer services support, messaging and many others.
At the AI Assistant Summit in San Francisco this January 25 & 26 Elena Sokolova, Data Scientist from Booking.com will discuss her work in creating personalised booking assistants to improve customer experience. In advance of her presentation at the summit we spoke with Elena about her current work and the progressions of AI at Booking.com.
How did you begin your work in AI, and more specifically in travel e-commerce?
"Six years ago when I was graduating from the Math department and was thinking about my future career I saw the first online course of Andrew Ng about machine learning. It was the time when the course had just launched and the Coursera website didn’t even exist. I took the course and it changed my life. I realized that AI is what I actually want to do and decided to do a PhD in the field. Being a travel addict I was using Booking.com every month during my PhD and was always curious about the AI that the company employ. I then met a data scientist from Booking.com at a conference, and was compelled to apply for a data scientist position. This brought me to a fascinating world of travel e-commerce."
What are the main challenges in implementing personalized assistance in travel bookings and why do they exist?
"Personalized assistance is a very challenging topic in itself and especially in Booking.com. One of the main challenges, is the continuous cold start problem (explained here). The majority of users browsing the website do so whilst not logged in, plus travelling is something that you do once, maybe twice a year, so we do not have a long shopping history of our users, in comparison, for example, to the websites like Amazon or Netflix. That makes it very challenging to build models based on user’s purchase history. On the other side, having millions of users per day we can make very good context-based models that take into account aspects such as user language, location, device and so on."
How are you using AI for a positive impact?
"AI is a great tool to understand what kind of information is relevant to our user and make search process more efficient. Different users have very different requirements, for example if you would like to stay for a weekend in Amsterdam for a business trip, you would like to stay in the hotel in the city center. However, if you are going for a week for a beach trip you might prefer an apartment with a sea view. We try to detect that needs from users’ search patterns and optimize the website for each user. We build our AI models and then test them in production in A/B testing to detect whether the change that we introduced actually helps our users. It is a great pleasure to actually see in real experiments that model that you developed helped millions of people to find a better place for their vacations."
What developments of AI are you most excited for, and which industries do you think will be most impacted?
"One of the greatest developments in the current AI is Automated Machine Learning. This is a method that allows you to find the best parameters for your algorithm automatically. We have already tried Auto ML and it does bring an improvement in accuracy and makes the life of a data scientist way easier."
There’s so much discussion around AI ‘stealing’ jobs - do you think this is the case, or will we actually see more jobs being created as a result of AI?
"I do not think that AI will steal our jobs in future. I do think that some jobs might become redundant but new jobs will arise, and overall AI will make our lives easier and more creative. The invention of the cars made the jobs of the coachman obsolete, but it introduced a new job of a driver, the same way AI will create new job opportunities for us in future."
For any company providing an online service whether that be travel, e-commerce, financial or a whole host of others, the demand for instant solutions to users frustrations and an ease of website usability is becoming imperative for companies to thrive. There are so many big players in these industries, and those who don’t employ AI as a means of optimising efficiency and providing a better user experience will get left behind.
Join us at the AI Assistant Summit in San Francisco this January 25 & 26 to learn from global pioneers and leading minds in the field to hear how they’re improving their results through AI. Confirmed speakers include Lionel Cordesses, Innovation Project Manager, Renault; Siva Reddy, Postdoc, Stanford NLP Group, Or Cohen, Head Of Engineering and AI, Mode.ai, Pararth Shah, Research Engineer, Google, Alison Darcy, Founder & CEO, Woebot & many more.