MemBuddy

Studying has never been so convenient.

Comment

Inspiration

Studying: a task we all love to hate. Whether it's hitting the books for that chemistry midterm, cranking out a few more math problems, or reviewing those vocab cards for the fifteenth time, the task is normally very unenjoyable, though the results are anything but. Oftentimes, the hardest part is just getting started; once you get in the groove, it's not too hard to keep on going, because each little success, whether it be correctly answering a study problem or getting another flash card right, motivates you. MemBuddy was created to address this very problem. By removing the first barrier toward learning, MemBuddy is able to help you become a better student while simultaneously making learning enjoyable and spontaneous.

Usage

MemBuddy is an SMS application that allows you, the user, to input different texts or questions. If you input a text, MemBuddy will generate questions for you using a NLP algorithm that converts information into pertinent questions. Note that you can even input pictures of textbooks; MemBuddy will automatically read the text and create questions for you. Afterward, you can choose to receive questions by texting MemBuddy "next," at which point MemBuddy will give you a question to answer. MemBuddy's distinguishing feature, however, is its periodic questioning. A few times each day, MemBuddy will text you a question for you to solve, assuming you have solved its previous question. As a result, MemBuddy is actively encouraging you to study by putting a problem in front of you, removing that initial barrier. MemBuddy is also capable of sorting questions into different personalizable classes, which you can ask for questions for if you feel that you want to study one particular topic.

How we got the idea

We first decided on the idea of MemBuddy by realizing that we wanted to find an effective way to spread out our studying, and study through a medium we always had with us. However, we didn't want to manually input all sorts of questions relating to our studies, and decided that generating questions was far more convenient. Combining this with Google's Vision API, we were able to simplify the process of making study material to just taking a picture. After finding an efficient algorithm online that generated questions from sentences, we felt that MemBuddy was a really cool idea, and went from there.

Challenges we ran into

One of our biggest challenges was creating an algorithm that effectively represented the questions stored in MemBuddy. We had to create a program that would emphasize questions that the user had previously gotten wrong, while also making sure that no questions were repeated frequently. In order to do so, we ended up assigning each question a variable known as adeptness, which was a rough measure of a user's knowledge of a question, and combined this with the last occurrence of the question to figure out how likely it was for a particular question to come up next. The algorithm we created worked very well, and we're very happy with how we turned one of our biggest challenges into a success.

What we're proud of

We're pretty proud of how we combined Google's Vision API with the NLP algorithm we found in order to generate questions from pictures and text. This is something that we felt would simplify the studying process immensely; by taking a picture of a textbook page, a user would have access to several questions that helped them understand the page. In addition, we like how we offered several options for the user, from the different methods of uploading to the different classes. We felt that doing so made MemBuddy more personalized to the user.

What we learned

Before this weekend, most of the team had no experience with computer vision or NLP, and only very limited to no experience with Python. Despite this, we were willing to learn new skills and technologies in order to build an app we felt passionate about, and we believe we built the best product we could given our background.

The future

We're unsure about MemBuddy's future, but we think that the three of us will probably be using it. Either way, we all had a lot of fun with this project.

Share this project:

Updates