EdNet: A Large-Scale Hierarchical Dataset in Education

Choi, Youngduck; Lee, Youngnam; Shin, Dongmin; Cho, Junghyun; Park, Seoyon; Lee, Seewoo; Baek, Jineon; Bae, Chan; Kim, Byungsoo; Heo, Jaewe

doi:10.1007/978-3-030-52240-7_13

Cited by 115 publications

(60 citation statements)

References 3 publications

(2 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…EdNet [3] is the largest publicly available benchmark dataset in education domain consisting of interaction logs collected by Santa 1 . We conduct experiments on an updated version of EdNet-KT1, which contains problem-solving logs from January 1st, 2019 to June 1st, 2020.…”

Section: Experiments 51 Datasetmentioning

confidence: 99%

See 1 more Smart Citation

SAINT+: Integrating Temporal Features for EdNet Correctness Prediction

Shin¹,

Shim²,

Yu³

et al. 2021

LAK21: 11th International Learning Analytics and Knowledge Conference

Self Cite

View full text Add to dashboard Cite

We propose SAINT+, a successor of SAINT which is a Transformer based knowledge tracing model that separately processes exercise information and student response information. Following the architecture of SAINT, SAINT+ has an encoder-decoder structure where the encoder applies self-attention layers to a stream of exercise embeddings, and the decoder alternately applies self-attention layers and encoder-decoder attention layers to streams of response embeddings and encoder output. Moreover, SAINT+ incorporates two temporal feature embeddings into the response embeddings: elapsed time, the time taken for a student to answer, and lag time, the time interval between adjacent learning activities. We empirically evaluate the effectiveness of SAINT+ on EdNet, the largest publicly available benchmark dataset in the education domain. Experimental results show that SAINT+ achieves state-of-the-art performance in knowledge tracing with an improvement of 1.25% in area under receiver operating characteristic curve compared to SAINT, the current state-of-the-art model in EdNet dataset. CCS CONCEPTS• Computing methodologies → Artificial intelligence; Neural networks; • Applied computing → Interactive learning environments; • Social and professional topics → Student assessment.

show abstract

Section: Experiments 51 Datasetmentioning

confidence: 99%

“…The strength of the DL based models has become amplified with the advent of large scale public benchmark dataset in Artificial Intelligence in Education. EdNet [3] is such a publicly available dataset which is the largest in scale with more than 131M learning activities records from around 780K students.…”

Section: Introductionmentioning

confidence: 99%

SAINT+: Integrating Temporal Features for EdNet Correctness Prediction

Shin¹,

Shim²,

Yu³

et al. 2021

LAK21: 11th International Learning Analytics and Knowledge Conference

Self Cite

View full text Add to dashboard Cite

show abstract

“…EdNet: A Large‐Scale Hierarchical Dataset in Education 49 is a new dataset described in Choi et al (2019) which has also an competition associated with it. The dataset is based on data collected over 2 years by Santa; a multi‐platform AI tutoring service with more than 780 K users in Korea available through Android, iOS and web.…”

Section: Edm Public Datasets Repositories and Data Sourcesmentioning

confidence: 99%

Review on publicly available datasets for educational data mining

Mihăescu

Popescu

2021

WIREs Data Min & Knowl

View full text Add to dashboard Cite

The availability of a dataset represents a critical component in educational data mining (EDM) pipelines. Once the dataset is at hand, the next steps within the research methodology regard proper research issue formulation, data analysis pipeline design and implementation and, finally, presentation of validation results. As the EDM research area is continuously growing due to the increasing number of available tools and technologies, one of the critical issues that constitute a bottleneck regards a properly documented review on publicly available datasets. This paper aims to present a succinct, yet informative, description of the most used publicly available data sources along with their associated EDM tasks, used algorithms, experimental results and main findings. We have found that there are three types of data sources: well‐known data sources, datasets used in EDM competitions and standalone EDM datasets. We conclude that the success of the future of EDM data sources will rely on their ability to manage proposed approaches and their experimental results as a dashboard of benchmarked runs. Under these circumstances, the reproducibility of data analysis pipelines and benchmarking of proposed algorithms becomes at hand for the research community such that progress in the EDM domain may be much more easily acquired. The most crucial outcome regards the possibility of continuously improving existing data analysis pipelines by tackling EDM tasks that rely on publicly available datasets and benchmarking data analysis pipelines that use open‐source implementations. This article is categorized under: Application Areas > Education and Learning Fundamental Concepts of Data and Knowledge > Big Data Mining

show abstract

“…Instead, they have been primarily applied to data generated in STEM learning contexts, such as math [5,11,43], algebra [37], and electrical engineering [19]. Only one dataset contains data generated in language learning contexts (preparation for an English as a secondary language test) [7], although it is unclear whether the tasks learners worked on in that dataset involved memorization tasks.…”

Section: Das3hmentioning

confidence: 99%

Linguistic Skill Modeling for Second Language Acquisition

Zylich

Lan

2021

LAK21: 11th International Learning Analytics and Knowledge Conference

View full text Add to dashboard Cite

To adapt materials for an individual learner, intelligent tutoring systems must estimate their knowledge or abilities. Depending on the content taught by the tutor, there have historically been different approaches to student modeling. Unlike common skill-based models used by math and science tutors, second language acquisition (SLA) tutors use memory-based models since there are many tasks involving memorization and retrieval, such as learning the meaning of a word in a second language. Based on estimated memory strengths provided by these memory-based models, SLA tutors are able to identify the optimal timing and content of retrieval practices for each learner to improve retention. In this work, we seek to determine whether skill-based models can be combined with memory-based models to improve student modeling and especially retrieval practice performance for SLA. In order to define skills in the context of SLA, we develop methods that can automatically extract multiple types of linguistic features from words. Using these features as skills, we apply skill-based models to a real-world SLA dataset. Our main findings are as follows. First, incorporating lexical features to represent individual words as skills in skill-based models outperforms existing memory-based models in terms of recall probability prediction. Second, incorporating additional morphological and syntactic features of each word via multiple-skill tagging of each word further improves the skill-based models. Third, incorporating semantic features, like word embeddings, to model similarities between words in a learner's practice history and their effects on memory also improves the models and appears to be a promising direction for future research.

show abstract

EdNet: A Large-Scale Hierarchical Dataset in Education

Cited by 115 publications

References 3 publications

SAINT+: Integrating Temporal Features for EdNet Correctness Prediction

SAINT+: Integrating Temporal Features for EdNet Correctness Prediction

Review on publicly available datasets for educational data mining

Linguistic Skill Modeling for Second Language Acquisition

Contact Info

Product

Resources

About