Inspiration
Match.com challenge
What it does
Creates a personality profile for pets based on descriptions entered by owners, then matches that profiles to others to help create romantic or platonic connections for people as well as potential pet playdate connections.
How we built it
We used a word-emotion lexicon to assign a numerical emotional score to adjectives entered by owners about their pets to create a holistic personality profile for the pet, then used Euclidian distance to measure the similarity between profiles. To analyze essay answers, we used machine learning algorithms from existing research on essay analysis for personality testing to determine additional information about both the subject of the essay (the pet) and the writer of the essay (the owner).
Challenges we ran into
We knew that machine learning was necessary for this project, but we didn't have the time or resources to train our own data, so we spent almost a quarter of the hackathon time researching solely to find pretrained data and machine learning algorithms we could implement in our project.
Accomplishments that we're proud of
We are proud of the sheer amount of work we were able to complete in the limited time frame. Our team had only two members and our project required extensive research before implementation. We are proud of ourselves for coming up with such a complex and roundabout way of determining human capability, then being able to bring that vision to reality in only 36 hours.
What we learned
We learned that Googling is a skill, and hackathons are determined less by experience and classroom intelligence than hard work and creativity. We also learned that hashing out the details of a good idea before you even touch an IDE to code helps the process run much more smoothly when the time comes to bring an idea to life. Although we used probably half of our time to research, that paid off in the end by making the development easier.
What's next for My Dog Ate my Dating Profile
We are working to make the web application ready for full deployment, and we would like to use a more complex method for profile matching, such as a machine learning algorithm, rather than the Euclidian distance formula that we utilized for the initial project showcase.
Log in or sign up for Devpost to join the conversation.