Our inspiration came from the online platform, Lunch Club. Lunch Club's AI and machine learning allows for users to connect on similar interests. We realized with the growing popularity of coffee chats that there is no platform like this that exists for women+ in technology to connect on their experiences.
What it does
According to ICT Solutions and Education, in 2018 women make up only a quarter of all jobs in the tech industry. In 2020, a study conducted by AnitaB.org Institute found that women now make up 28.8% of tech jobs. The lack of women in the industry leaves women leads to imposter syndrome and needing additional support from other women. This is why we created SheConnect. SheConnect is a website that uses machine learning and artificial intelligence to connect like minded women+ in tech. First, you create a profile and add in your interests and role. Then SheConnect will do the rest of the work for you, curating to find women in technology who also have the same interests. Users have the ability to book coffee chats with other women to provide each other support, motivation, and a safe space.
How we built it
For front-end, all screens were designed using React.
For the backend we used React, Node.Js, and MySQL. MySQL was used to safely store SheConnect's users' information, including their log in information and passwords. We used React and Node.js to implement the users' profiles and to creates matches for the users based on common interests that were written in their bios. The matching algorithm works by using user information to retrieve other profiles that have the same interests and are close in proximity to the user. Basically, we created a compatibility rating that suggested profiles based on different criteria ordered by importance: location, common interests, and how active the profile is. The more users on our app the more effective our algorithm would be, because of the larger dataset. To test our work we created user bios to work as a data set for our algorithm. We implemented machine learning using natural language processing using Python NLTK which tokenized the users' bios and created predictions based how similar users' profiles were to each other using Agglomerative Clustering from Sklearn. We also used Google Maps/Geolocator APIs for geolocation to be able for users to find other nearby users. The users also can upload and change their profile pictures.
Challenges we ran into
We ran into quite a few challenges throughout the past 24 hours. It was hard to find a way to bridge the gap between front end and the backend, but we managed to find common ground using React. None of us had worked with natural language processing before, but we were determined to incorporate it with our project. We put in a lot of research to figure out how to implement it.
Accomplishments that we're proud of
We are especially proud that we were able to use natural language processing to create clusters of users with common interests, in order to give recommended profiles. We created a really ambitious platform - and we are really proud of the fact that we were able to submit it as a project!
What we learned
We all learned so much during Technica! Our team was determined to learn about machine learning and implementing it into our project. With workshops and a lot of research, we were successful in learning how to make profile matches with AI to make user connections even better!
What's next for SheConnect
The next step for SheConnect is to further improve the AI so users can find the best matches to have coffee chats. A future feature we would like to implement is integrating a space where users can contribute to each other’s open source projects.