We constructed a program that allows a user to dynamically generate a vacation itinerary from the comfort of their own home by conversing with Alexa. This is revolutionary, because we show that vacation planning doesn't have to be stressful. Generating an itinerary without the hassle. That is our dream.

This project was inspired by the creative opportunities presented by the Amazon Echo's unique VUI. While trip planning skills have been made for Alexa, none have been made that combine extensive data mining and inference techniques that can help a user come up with exciting places to visit. Technological advances have led to an unnecessary increase of the level of complexity in the every day lives of many people. While technology is meant to make lives simpler, its users often face a steep learning curve. We aim to simplify the technicalities involved in vacation planning by using Amazon’s Alexa as a tool to facilitate the decision making process by giving vacation suggestions and helping users book those vacations. Inspired by Trip Advisor’s need to bring awareness to it’s flight and hotel booking services, we aimed to make the process as approachable as possible and to decrease the barrier to commitment.

What it does

Trip Planner helps people decide on and implement vacation plans, given an idea of what type of vacation they are seeking, their financial availability, and their temporal availability. It works by soliciting information on flights, hotel stays, tourism ratings, and activities of interest from various APIs, and testing the feasibility of various suggestions based on user requirements. In addition it takes user preferences for time of year and climate to generate better suggestions. Trip Planner is an interactive user interface that accessed multiple data sources to generate recommendations based on real time querying of available flights, hotel rooms, and climate information.

How we built it

We built it using a combination of NodeJS and Python. The python, hosted on the backend, ran a full set of data mining and api-accessing programs that assembled the required data based off user response. The frontend, build in NodeJS, detailed the interactions between the Alexa and the user. We used the TripAdvisor API to find locations of interest, while querying multiple other travel, lodging APIs with seasonal climate data for a given destination.

Challenges we ran into

Constructing a consistent workflow of questions for the user to provide relevant vacation recommendations based on budget, climate, duration, etc while obtaining and distributing enough information to perform API queries. Another major challenge was finding relevant data from multiple sources and simultaneously extracting and interfacing that data with the user demands to facilitate a smooth user experience through Alexa.

Accomplishments that we're proud of

Being able to query and leverage multiple data sources simultaneously while interacting with a user to generate recommendations is no simple feat. We were successful in generating a recommendation system using such a model, and hope to have positively impacted the lives of future vacations goers.

What we learned

Leveraging so many sources required an enormous amount of effort. This was no easy task when it came to coordinating the merger of multiple data streams to create one effective product. We also learned much about the complexities of real world human-computer interaction (HCI).

What's next for Trip Planner

There are many directions to take in the future. One option is to offer more services. Our original idea was to use the TripAdvisor API to book vacations, but TripAdvisor only offers such data to corporate business partners. With access to better API data, we can have Alexa proceed with the booking process, and reduce system latency. We also have visions of using more sophisticated statistical methods for suggesting vacation destinations based on user history.

Share this project: