Mr.Worldwide

Inspiration

When travelling in a new place, it is often the case that one doesn't have an adequate amount of mobile data to search for information they need.

What it does

Mr.Worldwide allows the user to send queries and receive responses regarding the weather, directions, news and translations in the form of sms and therefore without the need of any data.

How I built it

A natural language understanding model was built and trained with the use of Rasa nlu. This model has been trained to work as best possible with many variations of query styles to act as a chatbot. The queries are sent up to a server by sms with the twill API. A response is then sent back the same way to function as a chatbot.

Challenges I ran into

Implementing the Twilio API was a lot more time consuming than we assumed it would be. This was due to the fact that a virtual environment had to be set up and our connection to the server originally was not directly connecting.

Another challenge was providing the NLU model with adequate information to train on.

Accomplishments that I'm proud of

We are proud that our end result works as we intended it to.

What I learned

A lot about NLU models and implementing API's.

What's next for Mr.Worldwide

Potentially expanding the the scope of what services/information it can provide to the user.

Built With

• amazon-web-services
• dark-sky
• google-maps
• google-translate-api
• microsoft-bing-news-search-api
• python
• rasa-nlu
• twilio