When you visit the “ Prepare
for JEE – NEET – CET “ section of www.My-Teacher.in
, you come across following statement :
“ Over 294
Quintillion Unique tests. Yes , our
system can generate more than 294 Quintillion unique Test Papers from our
extensive question bank , ensuring no two tests are ever the same “
How do we conjure up this ( seemingly similar ) Magic ?
Our clever team member ( developer ) Kishan explains as follows :
“Here's
how the LLM's intelligence is leveraged:
When generating questions,
the LLM was instructed to determine the difficulty based on factors such as the
complexity of the concept, and the potential for confusion among students.
For example, simple recall
or basic concept questions
may be classified as 'Easy,'
questions requiring multi-step reasoning as 'Medium,'
and questions involving advanced concepts or tricky
interpretations as 'Hard “
Now , I was vaguely aware that Microsoft uses some similar process for
generating UNIQUE test papers for its online exams of MCSE . So I asked ChatGPT
to compare our feature with that of Microsoft . It came up with the following tabulation
( - a vindication of my request to NTA to adopt this technology for
administering NEET ) :
|
Aspect |
MCSE
Certification Tests |
|
|
Objective |
To generate unique tests
for students, ensuring consistent difficulty across different sets. |
To ensure test integrity
by providing unique questions with consistent difficulty across test-takers. |
|
Technology
Used |
Utilizes LLM (Large
Language Model) to determine question difficulty based on concept complexity
and confusion potential. |
Uses adaptive testing
technology and a large question pool to maintain consistent difficulty. |
|
Difficulty Determination |
Questions are classified
as 'Easy,' 'Medium,' or 'Hard' based on the complexity and reasoning
required. |
Questions are tailored
in difficulty, maintaining an overall balanced difficulty level across
different test versions. |
|
Test
Composition |
Each test includes 2
easy, 2 hard, and 6 medium-level questions, selected from the database. |
Each test is generated
from a large question pool, with questions dynamically selected to match the
required difficulty level. |
|
Purpose
of Question Variation |
Ensures that even though
all questions vary, the test difficulty remains consistent for all students. |
Prevents cheating by
ensuring two test-takers cannot receive identical questions, while
maintaining consistent difficulty. |
|
Scalability |
Designed to be scalable
for different students across various subjects and grades. |
Scalable for different
candidates globally across various certification tracks. |
Both systems aim to provide a fair and
consistent assessment experience by dynamically adjusting the test content to
maintain a consistent level of difficulty.
With Regards,
Hemen Parekh
www.My-Teacher.in / www.HemenParekh.ai / 04
Sept 2024
