1 / 11

Graph-based Adaptive Diagnosis

This CS 194 final project explores the implementation of a graph-based adaptive diagnosis system for student evaluations in math. The system responds and adapts to student answers, accurately assessing their understanding and identifying weak topics. It utilizes a directed graph to calculate student mastery and test confidence.

tylera
Download Presentation

Graph-based Adaptive Diagnosis

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Graph-based Adaptive Diagnosis CS 194 Final Project Results Jessica Wang Advisor: AmitSahai

  2. Problem: • Create a student evaluation system • More effective than a regular test • Responds and adapts to the student’s answers • Accurately assess student understanding • Identify the weak topics/areas from just a few questions

  3. Proposed Implementation • Utilization of a directed graph • Subtopics are represented by nodes • Edges have directions, weight, and type • Subject: math • Calculate student’s mastery in topic • Calculate test’s confidence of that estimation

  4. Actual Implementation • Graph of nodes and edges • Edges: traversal done between • Critical nodes • subtopics that are absolutely necessary for the next topic • Nodes • 2 types of nodes: critical and normal • Traversal starts out from critical nodes • Evaluation algorithm: 2 parts • Part 1: region determination • Part 2: region evaluation

  5. Region Exploration • Traversal algorithm between regions: • Start from the easiest region • Move to higher/lower region depending on student’s response • Return the region to be explored • Return whether this algorithm is conclusive/inconclusive

  6. Subtopics determination - Algorithm • Algorithm to explore region • Weird marker • Trickle-down estimation of mastery • Parallel traversal of routes: pick the next highest topic to ask

  7. Subtopics determination – Sample Region

  8. Issues • Ambiguous results: did student really master the region/topic? • How many questions to ask? • Per topic • Per test • How many points to deduct? • Are points too specific? • Use a more general marker

  9. Results • Determines the correct region most of the time • Algorithm sometimes underestimates student ability • Correctly takes into consideration student guessing • When answered incorrectly 20% of the time • Less accurate when student misses a higher percentage (33% of questions wrong)

  10. Results Like your normal math test! • Efficiency • Most efficient when everything is right • Worst case: around 30 questions • Average case: 15-20 questions (short questions) • Confidence of estimation • Does mastery marker match skill indication? • Did weird marker go off?

  11. Conclusion/Future Work • Evaluation system • Identifies the boundary between regions mastered and not yet mastered • If boundary is not clear, identifies a boundary within unclear region • Works relatively well: should have better confidence gauge • Give the test to actual middle/high school students • Implement graphic representation of the results

More Related