Binary Assessment Data Storage for Machine Learning Pattern Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional dichotomous assessment items face challenges in storing and processing student responses efficiently, leading to data loss and inconsistent processing, which limits the use of machine learning pattern detection and data analysis.
Innovation Solution
The system employs a non-dichotomous generative assessment approach that stores student answers in binary format without delimiters, allowing for concatenation and conversion into vector representations for machine learning, while using blockchain for immutable storage and histogram analysis to preserve data provenance and enhance processing reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional dichotomous assessment items store responses separately with delimiters, then data storage is straightforward, but data loss occurs and processing becomes inconsistent
Solution Approach 1:
The patent segments the response data into individual binary digits (0s and 1s) representing each answer choice, eliminating the need for delimiters and external metadata. Each response is stored as a continuous binary string where position encodes meaning, preventing data loss and ensuring consistent processing across all assessments.
Solution Approach 2:
The patent creates a binary copy of the response data where each answer choice is represented by a binary digit. This binary representation serves as a lossless copy that preserves all information without requiring additional storage for delimiters or formatting, thereby preventing data loss while maintaining processing consistency.
2Loss of information
If traditional dichotomous items are used, then assessment structure is simple, but information density is low requiring more time to administer
Solution Approach 1:
The patent merges multiple answer choices into a single response field where students can select multiple options simultaneously. The system combines these selections into a binary string representation, increasing information density without requiring multiple separate questions, thereby reducing administration time while preserving all response information.
Solution Approach 2:
The patent changes the parameter representation from categorical (A, B, C, D) to binary (0, 1), allowing multiple selections to be encoded in a compact format. This parameter transformation increases information density by enabling multiple states to be represented efficiently, reducing the time needed to administer assessments while maintaining full information capture.
3Quantity of substance
If temporarily stored response data is deleted after scoring, then storage space is freed, but data provenance is lost
Solution Approach 1:
The patent creates a permanent binary copy of the response data that is stored indefinitely. This binary representation serves as both the scoring data and the provenance record, eliminating the need to delete original responses while maintaining storage efficiency. The binary string preserves all information about student selections without requiring additional storage for metadata or delimiters.
Solution Approach 2:
The patent transforms the response data into a binary parameter representation that is both space-efficient and permanently retainable. This binary encoding reduces the storage footprint compared to traditional delimited text while preserving complete data provenance, allowing indefinite retention without compromising storage capacity.
4Extent of automation
If machine learning pattern detection is applied to traditional assessment data, then analysis capability is enhanced, but data processing becomes inconsistent
Solution Approach 1:
The patent transforms assessment response data into binary parameters that are uniformly structured and delimiter-free. This standardization creates consistent input data suitable for machine learning pattern detection, enhancing automation capability while ensuring processing consistency across all assessments. The binary representation eliminates variability in data formatting that previously caused inconsistent processing.
Data Source
AI summary
Disclosed herein are systems, media, and method for training machine learning algorithms to identify an existence of a characteristic, a probability of the existence of the characteristic, or both for each a plurality of students comprising collecting a first plurality and second plurality of concatenated answers, creating a first training set comprising the collected sets of concatenated answers, and training the machine learning algorithm in a first stage using the first training set.