The building of Mindset at TechCrunch Hackathon 2016
Using Natural Language Processing to discover new practice problems for teachers and students
I met Tom Goldenberg over a year ago at a ruby on rails meetup. We had a great conversation about a chess platform he was building and have met periodically since to talk about tech, business, and eventually forming a team for a hackathon.
Fastforward to the weekend of May 7–8th and we, plus Prem Ganeshkumar, Shabnum Gulati, and lowell mower met to build something for Techcrunch Disrupt NY. The competition — 645 other hackers & designers all fueled on red bull, snacks, hopes of winning some part of the $82k in prizes, and the dream of being annointed the competition champion.
We decided to focus on applications of Natural Language Processing (NLP) in education. We knew a few things: 1) students and teachers are facing more pressure for academic achievement, 2) students and teachers are more likely to posssess technology today, 3) the United States ranks mediocrely compared to worldwide academic achievement (35th in Math and 27th in Science*), and 4) if we could augment teachers’ capabilities we could help.
We ended up building: MindsetApp
Using Natural Language Processing we help teachers and students discover additional practice problems directly related to the topics they are studying. For example a teacher, Emma Morgan who is teaching Algebra 1, has used the same 10 practice problems for the topic of Polynomials for the last 5 years, but she’s always known some students want to augment their studies and she should probably swap out those problems either to 1) avoid students who copy answer from years prior, and/or 2) to improve/freshen the problems.
So we built a Natural Language Processing (NLP) powered platform bringing teachers and students crowdsourced and proprietary math problems.
Teachers, like Emma Morgan login using the HMH identity endpoint and then upload their syllabus.
Our NLP team used IBM Watson to build an engine that parses the syllabi extracting topic tags. The teacher can upload/input practice problems associated with those tags and then using those same tags, comb our database for related problems which she can select to use or share with students.
From the student end. They can use either the web or mobile app with the code provided by their teacher. This brings them directly to the practice problems of both their teacher’s design and those the teacher has selected from the crowdsourced database. All in all, this provides the students with access to much needed extra practice.
We started with Algebra 1 and 2 plus Geometry but see uses cases in additional mathematics classes and even beyond.
We add to your problems and are proud of it.
Overall this competition was a lot of fun. It was pretty amazing seeing what our team built in such a short time. Truly impressed by the skillsets each person brought to the table. It was also great to see the amazing products built by all the others — scroll down to see more on that. Loved the spirit of this event and its competitors. Looking forward to the next one.
Check out the video of Tom pitching the app:
Interested in seeing the complete field of competitors?
Dr Speech aims to help people who speak English as a 2nd language to improve the accuracy of their pronunciation.disruptny2016.devpost.com
Of course, hat tip to the overall winners, AlexaSite.
Lilwil's personalized learning engine teaches teachers how to teach It's been a long night at the Brooklyn Cruise…techcrunch.com