Travel, expense, and machine learning
"Artificial Intelligence and Deep Learning could be as profound and maybe even bigger than the shift to mobile and cloud" – Frank Chen (Andreessen Horowitz)
"Google has moved from a search first to a mobile first to an AI first company" – Google CEO Sundar Pichai
“Humans [will] increasingly work side by side with robots, software agents and other machines.” – J.P. Gownder, Analyst | Forrester Research
So, it looks like artificial intelligence and machine learning could be the next big wave in computing; and could make a significant impact in our everyday lives.
Artificial intelligence represents computers simulating human intelligence (learning, problem solving) and machine learning describes a subset of AI focused on statistical algorithms that learn from data to make improved predictions. (Deep Learning is a further evolution of machine learning using layers of artificial neurons with large amounts of data.)
With its potential impact, many companies are talking about Machine Learning. For SAP, the entire portfolio of intelligent applications and services is called Leonardo Machine Learning.
Here at Concur, machine learning will certainly be increasingly important in the future - however part of the future has already arrived.
You can see SAP Leonardo in action at Concur as there are already services that have begun to utilize machine learning and other related data science algorithms to help simplify travel and expense for our customers. These services may point to areas where this growing field will continue to impact travel and expense in the future.
Taming the Wily Receipt – One Step towards Effortless Expenses
Everyone loves getting a receipt and then having to attach it to an expense report without misplacing it first. At Concur we let you take a picture of a receipt and digitally store it for you so you don’t lose it. We also use OCR (Optical Character Recognition) to extract the text from a receipt image and use machine learning to identify important fields such as total amount, date, currency, expense type etc. Using these fields, we then create an expense entry for you, which can be added to an expense report.
Similarly in Concur Expense for Outlook (a beta from ConcurLabs) Concur data science algorithms make it easier to submit an expense from directly inside of Outlook by helping pre-populate the expense entry with information it grabs from an electronic receipt that you have received via email .
Hotel Recommendations - Your ML Buddy Suggesting a Place to Stay
One important capability in the travel and expense world is providing recommendations. Travel applications can provide recommendations during the travel booking process based on your activities and the activities of others. For example, hotel recommendation in Concur mobile use data that helps you see that “Your co-workers have stayed here" Whether this statistical data analysis is strictly considered machine learning, one could imagine recommendations in the travel and expense world evolving as machine learning techniques evolve.
BOTs / Conversational Interfaces – A Virtual Travel Assistant
A Chatbot or Bot is a program that creates a conversational interface by using machine learning for natural language processing. A Bot tries to understand and respond to a conversational request made by an end user via text or speech.
For example, Hello Hipmunk is a virtual travel assistant which both provides travel recommendations and enables travel booking. One of the most popular travel chatbots in the US, it brings its conversational interface to applications such as email, calendar, Facebook Messenger, Skype, and Slack.
Fast Company described it as: “A dash of artificial intelligence [that] lets you search for flights and hotels without actually searching for flights and hotels.”
Similarly, Concur Lab’s “Concur for Slack” helps simplify travel and expense by bringing our services to where our users spend their work day. While in Slack, users can see their travel itinerary, upload receipts for expensing, create a quick expense, and view a summary of expense reports.
And, the “Concur for Alexa” prototype has similar capabilities using Amazon Alex’s voice interface.
Sentiment Analysis & Machine Learning Uncovering Customer Needs
Another recent project from Concur Labs showed how sentiment analysis can aggregate and assess ratings and reviews in an effort to improve end user satisfaction. Sentiment analysis uses natural language processing (another branch of AI) to convert written opinions about a product into a numerical sentiment rating. This makes automated analysis possible including using machine learning to identify trends over time.
This is just the beginning of how machine learning will impact the world of travel and expense, so stay tuned …..