Javascript must be enabled for the correct page display

Optimal word pair learning in the short term: using an activation based spacing model

Woudenberg, M.H. van (2008) Optimal word pair learning in the short term: using an activation based spacing model. Master's Thesis / Essay, Artificial Intelligence.

[img]
Preview
Text
AI-MAI-2008-M.H.van.Woudenberg.pdf - Published Version

Download (406kB) | Preview

Abstract

What would you do if you only had 15 minutes to learn a list of word pairs for an exam tomorrow? What learning strategy would you use? And given the short amount of time to learn, does the strategy even matter? These are questions addressed in this thesis. The motivation for this project was to find an adaptive, optimal learning schedule, that is proven to work in a real-life setting. Although a lot of work has been done on optimal learning paradigms, most of this research has focused on longer learning periods (> 30 minutes), and longer retention intervals (> 1 week). This is in stark contrast to informal reports on how students learn word pairs, which is more typically described as consisting of a single, shorter learning episode (< 30 minutes) one day before a test. To construct a optimal learning schedule, ACT-R’s spacing model (Pavlik and Anderson, 2005) is used to assess the internal representation of presented word-pairs. On the basis of this model a Dynamic Spacing method is constructed that repeats word pairs just before they are forgotten. We compared three variants of this method with a standard learning schedule. The three variants differed in the amount of adaptation to the individual’s behavior. Students (selected from 3 HAVO/VWO) were presented with a learning session of 15 minutes on Day 1, and got an unexpected exam the next day. Analysis of the results shows that learning word pairs in the Dynamic Spacing condition results in better scores, given that the most sensitive adaptation method is chosen. This improvement is strongest in those students with below-average skills in the tested domain. Although some issues remain, the work presented in this thesis shows that selecting an optimal Dynamic Spacing learning strategy improves the average results on the test by 10%, largely because students with below-average results score remarkably higher with the optimized learning schedule.

Item Type: Thesis (Master's Thesis / Essay)
Degree programme: Artificial Intelligence
Thesis type: Master's Thesis / Essay
Language: English
Date Deposited: 15 Feb 2018 07:28
Last Modified: 15 Feb 2018 07:28
URI: http://fse.studenttheses.ub.rug.nl/id/eprint/8463

Actions (login required)

View Item View Item