The goal of the Microsoft's challenge is to use AI for Good. Our inspiration comes from all our elderly relatives we know, who has a lot of time, because they aren't anymore as busy as we are with all our study and job stuff to do. But we don't want them to be lonely, so we invent our solution for our grannies and grandpas.
What it does
Our Vision is to connect People, especially elderly people who'd like to talk to someone about the Good old Times.
Through an easy way to talk to/meet like-minded people we aspire a opportunity to be independent and find new friends. It's all about bringing people together and create a possibility even for those who can't travel that good anymore.
Our society is growing older and older. Narturally there's a lot of elderly people who aren't as connected as they've been, especially when they don't have children. When retirement comes around you might loose some great connections from work, if you have a family they might also be busy and friends might just become rare. And in a retirement home you might not meet a lot of new people.
So that's what we aspire to change. We want you to meet or talk to new people who might become friends. We want you to be able to talk to like-minded people and enjoy remembering the Good old Times.
We don't want you to be alone or feel lonely.
Since many elderly people aren't quite used to the always changing technologies of todays society, the goal is to create an easy way of using modern technology by combining it with phones.
You can call and the first time you have to create your profile. But not like a never ending formular of fields you have to type into. Instead you just call us and you're talking to a speech-to-text/text-to-speech service. The service writes down your profile and saves it. Afterwards the AI matches registered users based on location, age, gender and interests and transfers your call to this number.
How I built it
The Frontend is a Webapplication which sents registration informations to the Backend and gets the generated recommended elder people to connect with. The Backend is built with Flask and Python. It implements an self-made machine learning model for clustering all the data. The model is built with the help of the powerful python library SciPy. The whole system runs in a docker container on the Azure platform and also uses as storage the Azure Cosmos DB (MongoDB).
Challenges I ran into
- deciding which machine learning method to use
- trouble with python versions
- docker tagging and updating in azure
- inexpierince with all the technologies we used
- to less time, to less sleep
- no frozen yogurth on the first day
- css doesn't want as we wanted
Accomplishments that I'm proud of
- used many technologies we never used before and it worked
- get to know the azure platform
- enjoy the time at the hackathon
- good css
- first machine learning thing ever, first python thing ever
What I learned
- crazy stuff about all the new technologies we used
- how to sort all balls in a ball pool
What's next for Good old times
- get the speech-to-text recognition to work
- call a api to start phone calls between matched pairs
- improve clustering algorithm