The hive is everywhere. Everyone has opinions. And anyone could look for advice to everyday questions. In fact, we make an average of 35,000 decisions a day. We’re connecting users all around the world so that any user can solicit opinions from anyone anywhere any time, in real-time. Just hive it on hiveit.org.
“Self-expression is everything. But I calibrate how I express myself based on how others respond.” Daniel, 25, Sao Paulo, Brazil.
“I only text a few people, but put out memes on the internet to random people all the time” Megan, 20, Long Island, USA
“I always poll my friends before I decide on an Instagram caption - it helps make sure I get more likes!” Kristin, 24, Berlin, Germany.
What it does
What if you could tap into a readily-available online community & collect suggestions on these everyday decisions in real-time?
And what if we could collect this data for a ML algorithm (fed onto a Messenger Chatbot or a voice assistant, like FB Aloha) that is enabled to make these simple everyday questions for you in a considered, human-centered way?
How I built it
We focussed on one use case: deciding what outfit to wear.
Built two modules:
Sourcing preference data through a fun user-friendly product that was designed using various UX tools and then put together on Wix and using JS.
Developed a ML algorithm that takes this user preference data & predicts preferences for future uploads.
The ML Algorithm was built as follows:
- Using Google cloud automated machine learning vision tool to classify clothes.
- A recommendation to a product will be assigned based on the popularity in the market, people will answer the question by selecting the recommended item & a percentage voting weight will be assigned to each item
- Using the output from stage 1, the selection process will be determined by applying a mathematical model we developed.
Challenges I ran into
-Initial challenges integrating our ML algorithms to the Wix Code since we were using Python. Transferred this over to node.js. -Design issues around the mobile site, which we were able to fix by implementing changes on the mobile version on Wix.
Accomplishments that I'm proud of
-Designing & building something that was usable and, in our opinion, very practical!
-There was a steep learning curve since all the platforms were very new to us. We're very proud of how we were able to pick them up together as a team.
What I learned
-Using Wix Code and integrating it with various back-end systems. -It was our first attempt at building a complex ML algorithm -Working with a very diverse team from all around the world, with a broad range of skills - this was the best thing we could learn!
What's next for HiveIt.org
- Show prototype to sample of users to collect initial feedback
- Test beyond clothing options
- Build out results page with demographics info (e.g. a map with where votes are coming in from using here.com)
- Define the monetization model -charge for app installs -trial & in-app purchase -freemium opportunities -Ripple payment per Q and payment per swipe