Alleviating the Cold Start Problem in Adaptive Learning using Data-Driven Difficulty Estimates

An adaptive learning system offers a digital learning environment that adjusts itself to the individual learner and learning material. By refining its internal model of the learner and material over time, such a system continually improves its ability to present appropriate exercises that maximise learning gains. In many cases, there is an initial mismatch between the internal model and the learner’s actual performance on the presented items, causing a ‘cold start’ during which the system is poorly adjusted to the situation. In this study, we implemented several strategies for mitigating this cold start problem in an adaptive fact learning system and experimentally tested their effect on learning performance. The strategies included predicting difficulty for individual learner-fact pairs, individual learners, individual facts, and the set of facts as a whole. We found that cold start mitigation improved learning outcomes, provided that there was sufficient variability in the difficulty of the study material. Informed individualised predictions allowed the system to schedule learners’ study time more effectively, leading to an increase in response accuracy during the learning session as well as improved retention of the studied items afterwards. Our findings show that addressing the cold start problem in adaptive learning systems can have a real impact on learning outcomes. We expect this to be particularly valuable in real-world educational settings with large individual differences between learners and highly diverse materials.

Download Full-text

Alleviating the Cold Start Problem in Adaptive Learning using Data-Driven Difficulty Estimates

Computational Brain & Behavior ◽

10.1007/s42113-021-00101-6 ◽

2021 ◽

Cited By ~ 1

Author(s):

Maarten van der Velde ◽

Florian Sense ◽

Jelmer Borst ◽

Hedderik van Rijn

Keyword(s):

Learning Outcomes ◽

Adaptive Learning ◽

Internal Model ◽

Cold Start ◽

Learning System ◽

Learning Performance ◽

Learning Material ◽

Actual Performance ◽

Individual Learner ◽

Cold Start Problem

AbstractAn adaptive learning system offers a digital learning environment that adjusts itself to the individual learner and learning material. By refining its internal model of the learner and material over time, such a system continually improves its ability to present appropriate exercises that maximise learning gains. In many cases, there is an initial mismatch between the internal model and the learner’s actual performance on the presented items, causing a “cold start” during which the system is poorly adjusted to the situation. In this study, we implemented several strategies for mitigating this cold start problem in an adaptive fact learning system and experimentally tested their effect on learning performance. The strategies included predicting difficulty for individual learner-fact pairs, individual learners, individual facts, and the set of facts as a whole. We found that cold start mitigation improved learning outcomes, provided that there was sufficient variability in the difficulty of the study material. Informed individualised predictions allowed the system to schedule learners’ study time more effectively, leading to an increase in response accuracy during the learning session as well as improved retention of the studied items afterwards. Our findings show that addressing the cold start problem in adaptive learning systems can have a real impact on learning outcomes. We expect this to be particularly valuable in real-world educational settings with large individual differences between learners and highly diverse materials.

Download Full-text

SCIENCE EDUCATION ADAPTIVE LEARNING SYSTEM AS A COMPUTER-BASED SCIENCE LEARNING WITH LEARNING STYLE VARIATIONS

Journal of Baltic Science Education ◽

10.33225/jbse/18.17.711 ◽

2018 ◽

Vol 17 (4) ◽

pp. 711-727 ◽

Cited By ~ 1

Author(s):

Zulfiani Zulfiani ◽

Iwan Permana Suwarna ◽

Sujiyo Miranto

Keyword(s):

Learning Styles ◽

Learning Style ◽

Adaptive Learning ◽

Science Learning ◽

Learning System ◽

Learning Material ◽

Learning Approaches ◽

Slow Learners ◽

Computer Based ◽

Adaptive Learning System

Students with their different learning styles also have their own different learning approaches, and teachers cannot simultaneously facilitate them all. Teachers’ limitation in serving all students’ learning styles can be anticipated by the use of computer-based instructions. This research aims to develop ScEd-Adaptive Learning System (ScEd-ASL) as a computer-based science learning media by accommodating students’ learning style variations. The research method used is a mixed method at junior high schools in Tangerang Selatan. The final product of the research is a special learning media appropriate to students’ visual, aural, read/write and kinesthetic learning styles. The uniqueness of the media is its form of integrated science materials, accommodating fast and slow learners, and appropriate to their learning styles. ScEd-Adaptive Learning System as a developed computer-based science learning media was declared as good and valid by four media experts and five learning material experts. ScEd-ALS for kinesthetic style has a high effectivity to improve students learning mastery (100%), consecutively aural (63%), read/write (55%), and visual (20%). This media development can be continued with the Android version or iOS to make it more operationally practical. Keywords: adaptive learning system, science learning media, computer-based instruction, learning style.

Download Full-text

Comparing the Effects of Stroke-Appearing and Stroke-Disappearing on Learning the Order of Strokes in Chinese Characters

Frontiers in Psychology ◽

10.3389/fpsyg.2021.704457 ◽

2021 ◽

Vol 12 ◽

Author(s):

Jon-Chao Hong ◽

Kai-Hsin Tai ◽

Ming-Yueh Hwang ◽

Pei-Hsin Lin

Keyword(s):

Learning Outcomes ◽

Learning System ◽

Learning Performance ◽

Technology Enhanced Learning ◽

Chinese Characters ◽

School Students ◽

Eeg Data ◽

Significant Difference ◽

E Learning ◽

Writing Chinese

Different approaches to stimulating perceptions in learning can be easily designed with technology-enhanced learning systems. This study aimed to explore how different approaches can influence learners' perceptions that may negatively or positively affect their learning performance of writing Chinese characters using the correct Chinese order of strokes (COS). We therefore designed an e-learning system which was subdivided into two modes: stroke-appearing (i.e., using red to mark incorrect strokes) and stroke-disappearing (i.e., using blanks to mark incorrect strokes) to indicate strokes written in the incorrect order. We then investigated the modes that would facilitate a higher level of attention and better learning outcomes. A total of 10 third-grade elementary school students participated in the experiment, divided into two test groups. Their EEG data were collected, and time series analysis and t-tests were utilized to analyze the differences. The results indicated that: (1) there was a significant difference in the attention levels of the students practicing with the stroke-appearing and stroke-disappearing modes when learning COS, and (2) there was a significant difference in the learning outcomes of the students practicing with the stroke-appearing and stroke-disappearing modes when learning COS. These findings support the specific role of stroke order knowledge in learning Chinese characters and the need for the design of an effective method for teaching children to learn Chinese characters.

Download Full-text

Continuous improvement of deaf student learning outcomes based on an adaptive learning system and an Academic Advisor Agent

Computers in Human Behavior ◽

10.1016/j.chb.2017.07.006 ◽

2019 ◽

Vol 92 ◽

pp. 536-546 ◽

Cited By ~ 5

Author(s):

Salah Hammami ◽

Fahman Saeed ◽

Hassan Mathkour ◽

Mohammed Amer Arafah

Keyword(s):

Student Learning ◽

Learning Outcomes ◽

Continuous Improvement ◽

Adaptive Learning ◽

Student Learning Outcomes ◽

Learning System ◽

Academic Advisor ◽

Deaf Student ◽

Adaptive Learning System

Download Full-text

Design Smart Games with requirements, generate them with a Click, and revise them with a GUIs

ADCAIJ ADVANCES IN DISTRIBUTED COMPUTING AND ARTIFICIAL INTELLIGENCE JOURNAL ◽

10.14201/adcaij20121315568 ◽

2013 ◽

Vol 1 (3) ◽

pp. 55-68 ◽

Cited By ~ 1

Author(s):

Vincenza Cofini ◽

Fernando De La Prieta ◽

Tania Di Mascio ◽

Rosella Gennari ◽

Pierpaolo Vittorini

Keyword(s):

Adaptive Learning ◽

Learning System ◽

Story Comprehension ◽

Learning Material ◽

Generation Process ◽

Inference Making ◽

Automated Generation ◽

Poor Comprehenders ◽

Adaptive Learning System

TERENCE is an FP7 ICT European project that is developing an adaptive learning system for supporting poor comprehenders and their educators. Its learning material are books of stories and games. The games are specialised into smart games, which stimulate inference-making for story comprehension, and relaxing games, which stimulate visual perception and not story comprehension. The paper focuses on smart games. It first describes the TERENCE system architecture, thus delves into the design of smart games starting from the requirements and their automated generation, by highlighting the role of the reasoning module therein. Finally, it outlines the manual revision of the generated smart games, and ends with short conclusions about the planned improvements on the automated generation process.

Download Full-text

A Service Based Adaptive U-Learning System Using UX

The Scientific World JOURNAL ◽

10.1155/2014/109435 ◽

2014 ◽

Vol 2014 ◽

pp. 1-9 ◽

Cited By ~ 1

Author(s):

Hwa-Young Jeong ◽

Gangman Yi

Keyword(s):

Cloud Computing ◽

Ubiquitous Computing ◽

Learning Style ◽

High Performance ◽

New Technology ◽

Learning System ◽

Learning Performance ◽

Learning Material ◽

Computing Environment ◽

Application Systems

In recent years, traditional development techniques for e-learning systems have been changing to become more convenient and efficient. One new technology in the development of application systems includes both cloud and ubiquitous computing. Cloud computing can support learning system processes by using services while ubiquitous computing can provide system operation and management via a high performance technical process and network. In the cloud computing environment, a learning service application can provide a business module or process to the user via the internet. This research focuses on providing the learning material and processes of courses by learning units using the services in a ubiquitous computing environment. And we also investigate functions that support users’ tailored materials according to their learning style. That is, we analyzed the user’s data and their characteristics in accordance with their user experience. We subsequently applied the learning process to fit on their learning performance and preferences. Finally, we demonstrate how the proposed system outperforms learning effects to learners better than existing techniques.

Download Full-text

Modelling The Learner Model Based Ontology In Adaptive Learning Environment

Journal of Disruptive Learning Innovation (JODLI) ◽

10.17977/um072v1i12019p34-45 ◽

2019 ◽

Vol 1 (1) ◽

pp. 34

Author(s):

Saida Ulfa ◽

Deddy Barnabas Lasfeto ◽

Citra Kurniawan

Keyword(s):

Learning Styles ◽

Adaptive Learning ◽

Domain Knowledge ◽

Personalized Learning ◽

Learning System ◽

Learning Performance ◽

Learning Goals ◽

Online Learners ◽

Learner Model ◽

Adaptive Learning System

Currently, the online learners are increasingly demanding more personalized learning since the web technology, and the learners have individual features of characteristics such as learning goals, experiences, interests, personality traits, learning styles, learning activities, and prior knowledge. A personalized learning process requires an adaptive learning system (ALS). In order to adapt, a learner model is required. Thus, modelling the learner model in an adaptive system environment is a key point to success in recommending the learner. The ontology-based approach was used to model the adaptive learning model in this research. Ontology is a graph structure that consists of a collection of contexts, relationships, and models which related to contexts. The ontology of the learner model enables to produce a description of learner’s properties which contains important information about domain knowledge, learning performance, interests, preference, goal, tasks, and personal traits.Keywords - Personalized Learning, Adaptive Learning System, Ontology, Learner Model

Download Full-text

RANCANG BANGUN MEDIA PEMBELAJARAN MULTIMEDIA INTERAKTIF PADA MATA PELAJARAN SIMULASI DIGITAL DI SMK

Voteteknika (Vocational Teknik Elektronika dan Informatika) ◽

10.24036/voteteknika.v7i1.103871 ◽

2019 ◽

Vol 7 (1) ◽

pp. 200

Author(s):

Syawaluddin Syawaluddin ◽

Elfi Tasrif

Keyword(s):

Learning Outcomes ◽

Interactive Multimedia ◽

Multimedia Learning ◽

Interactive Media ◽

Learning System ◽

Learning Material ◽

New Learning ◽

The Media ◽

Media Learning ◽

Almost All

Today the use of interactive media to support the learning system so that there is an increase in students towards learning outcomes. Technical in implementing learning media can involve almost all the senses of students. Because the media that has been designed is one of the new learning innovations, combined in it there are videos, images, sounds, and other aspects that support to explain the material so that students are easy to take part in learning. In this design there are several elements that are combined including images, animation, sound, and navigation buttons to facilitate users in using interactive media, this interactive multimedia learning media designed with an attractive appearance is presented in interactive CD so students can learn learning material independently. Keywords: Designing, Interactive Media Learning.

Download Full-text

The Design of Learning Material for Poor Comprehenders: Lessons Learnt from Experts

Jurnal Teknologi ◽

10.11113/jt.v63.1953 ◽

2013 ◽

Vol 63 (3) ◽

Author(s):

Vincenza Cofinia ◽

Fernando De la Prieta ◽

Tania Di Mascio ◽

Rosella Gennari ◽

Ivana Marenzi ◽

...

Keyword(s):

Artificial Intelligence ◽

Adaptive Learning ◽

Text Comprehension ◽

Learning System ◽

Learning Material ◽

Or Education ◽

Domain Experts ◽

Lessons Learnt ◽

Poor Comprehenders ◽

Adaptive Learning System

TERENCE is an FP7 ICT European project that is developing an adaptive learning system for poor comprehenders and their educators. The learning material is made of stories and smart games for stimulating reading comprehension. The design of stories and smart games is also based on data collected from experts for the analysis of the context of use of the system, and is incrementally revised via evaluations of prototypes of stories and games, with domain experts of text comprehension or education as participants. In particular, since smart games are semi-automatically generated via artificial intelligence technologies, they contain mistakes that have to be fixed by experts of pedagogy before the games are given to learners. In this paper we focus on the design and evaluations of the TERENCE stories and smart games for poor comprehenders via lessons learnt with domain experts.

Download Full-text