Berri

A personalized memory-retention tool for mathematical problem-solving.

User Interface

• implement core game design principles that primarily optimize for maintaining user flow during a state of solving math problems
• optimize for recurrent feedback loops for user's learning and performance metrics to maintain engagement in respect to frictionless app usability
• maximize data visibility and user retention to optimize user / student performance metrics, and further optimize internal algorithm assuming productionization and scale, similar to Brainscape's Confidence-Based Repetition Algorithm, but in respect to optimizing for algorithm-style questions that apply to mathematics.
• To view a live demonstration of the features, click on the video below.

Demonstration

Components

/tutors

• render tutor profiles in grid, allow students / users to filter by category & availability
• allow users to select and view tutor_profiles and join freemium sessions (enable multi-sided platform for educators, students, and tutors)
• store profile metadata and allow tutors to rank on the berri platform and gain traffic given their engagement and value creation for students.

/modules

• store a comprehensive set of subject categories which can handle LaTeX equations for higher-level mathematics courses beyond high school
• Store problem sets that were hardcoded with a mixture of problems varied by difficulty
• store confidence_scores given by the user for each problem they solve with respect to maintaining their flow without hitting maximum point of interruption that'd break their train of focus (attention span)
• store session_time, confidence_score for each problem they solve, track problem_completion, accuracy_instance_count (number of questions they solve correctly in a problem set)
• compute metrics that were converted and stored in pandas.dataframe at the end of each user session for the problem_set
• display short term performance metrics (specific to problem_set they complete)
• user will access math training gateway for problem_sets
• users can practice with currently existing problem_sets
• users will soon be able to add in their own problem sets from their math classes to offer a more personalized training experience in respect to the problem types they come across in their school, which varies between other districts and states

/dashboard

• render graphs, charts (data visualization) of the user's performance metrics / user analytics / for total # of questions solved correctly, est_mastery_time, user_xp (computed from # of correctly solved questions in respect to session_time)
• display graphs that track their performance with hardcoded metadata (category, difficulty) and display trends in respect to their accuracy
• graph more longer term trends and metrics that allow the user to view their performance in a clear manner with the problem sets they are doing in their mathematics class

/multiplayer

• real-time competitive gateway for online matches
• users will soon access marketplace to allow students / educators to create their own problem sets that can be shared with other students from different schools and grades
• users will soon be able to join & create public / private matches either locally within their classrooms, districts, and with other schools

/login && /signup

• enable google user authentication
• autogenerate profile for user

System Architecture

• Standard MVC (Model-View-Controller)

Algorithms

• Confidence-Based Repetition
• Object-Relational Mapping

Data Structures

• Pandas.Dataframe
• Nested JSON Objects
• Python Dictionary

Future Work

• randomization for problems imported from OCR or direct user input to automatically simulate test conditions
• public / private matches with keys generated from invite codes
• simulating test conditions for users via CRUD setup for problem sets (OCR trained model to identify question to prevent friction, user manually adds in multiple choice answers