[Pycon] [new paper] "Prabh Simran Baweja" - The Future of Chatbots in the Travel Industry

[info a pycon.it]

Title: The Future of Chatbots in the Travel Industry
Duration: 45 (includes Q&A)
Q&A Session: 15
Language: en
Type: Talk

Abstract: The Future of ChatBots in the Travel Industry

Introduction
Imagine the future where an intelligent assistant suggests you places you could visit on the weekend. And plans the whole trip for you as well. The bot already knows your preferences and previous experiences and makes you an offer you cannot refuse. Won't that be wonderful? This is the near future, my friends.

Everyone who loves to travel wants to have a smooth travel experience. ChatBots are trying to serve the needs of the users, coming up with content personalized according to each user. Wouldn’t you want to search for a hotel on the fly without setting up a plan to find the perfect getaway hotel for you? What if the Intelligent Assistant already knows what you like and comes up with exactly the same thing that you would love. That too, in a conversation that would last around 3-5 minutes, on your smartphone. ChatBots in the travel industry is changing our overall planning phase of the trip. I believe natural interfaces will play a major role in e-commerce.


Why Me?
I work at Booking.com in the ChatBot team, and I would like to share some insights, learnings, and feedback from our users over the past few months. The difference in the experiences of users when talking to the ChatBot has been striking. There have been a lot of cases where the ChatBot understood each and every part of the conversation, and cases, where the users were the ChatBot, misunderstood the message a user was trying to convey.  


How does it work? 
There are a lot of stages that are involved in processing the sentence that the user says, and coming up with an intelligent reply: 
The intent of the message (Book a hotel, Customer Support, Filters, Sorters, etc.)
Keeping in mind the previous conversations. 
Understand if the user is conveying something particular (Number of people in the trip, Number of days, Currency, Exact dates, Not sure about the dates etc.)
Figure out when the user comes back after a certain amount of time if they are starting a new conversation, or continuing the same one. 
These are some of the parts that need to take care of before coming up with a reply. Each of these parts is handled separately. 
There are two major parts of the ChatBot: 
The brain of the Bot: Talks to the user, identifies if it can reply directly, or needs Natural Language Understanding to parse the intent of the user, and then come up with a smart and relevant reply. 
NLU (Natural Language Understanding) : This part of the bot parses the sentences and comes up with intents in the message with some confidence. 


Why is this important? 
Mobile App time Spent Grows 69% Year-Over-Year.
There are ~1.3 billion active users on Facebook. ChatBots have the potential to impact their lives and help them with their travel needs.
Chat outperforms voice in user tests as people perceive chat to be more natural, faster, and more intelligent. 

How will this be useful for you? 
I will show all the components that are required for designing a chatbot. 
As a developer, you will know how to start building a bot after this talk. I will show bits and pieces of code which will be helpful for you to start. 
For the rest of the people, you will have a better understanding of how chatbots are changing our day to day lives and will be playing an important role in the future. 


Tags: [u'bigdata', u'machine-learning', u'django', u'Python']