SingPAD: A Knowledge Tracing Dataset Based on Music Performance Assessment
Saved in:
| Title: | SingPAD: A Knowledge Tracing Dataset Based on Music Performance Assessment |
|---|---|
| Language: | English |
| Authors: | Ying Zhang, Yan Zhang, Wei Xu, Zhifeng Wang, Jianwen Sun |
| Source: | International Educational Data Mining Society. 2024. |
| Availability: | International Educational Data Mining Society. e-mail: admin@educationaldatamining.org; Web site: https://educationaldatamining.org/conferences/ |
| Peer Reviewed: | Y |
| Page Count: | 9 |
| Publication Date: | 2024 |
| Document Type: | Speeches/Meeting Papers Reports - Research |
| Descriptors: | Artificial Intelligence, Intelligent Tutoring Systems, Knowledge Level, Music Education, Singing, Student Evaluation, Music Reading, Technology Uses in Education, Foreign Countries |
| Geographic Terms: | China |
| Abstract: | Knowledge tracing (KT) aims to model a learner's knowledge mastery level through his historical exercise records to predict future learning performance. Using this technology, learners can get appropriate customized exercises based on their current knowledge states, and thus the great potential of personalized teaching services such as intelligent tutoring systems and learning material recommendations can be stimulated. Currently, the mainstream datasets in KT include ASSISTments, EdNet, STATICS2011, etc., which are mainly based on objective testing data in the fields of mathematics and language, lacking of datasets on music performance assessment. Therefore, based on the context of performance assessment in music education, specifically sight-singing evaluation, we introduce "SingPAD," the first dataset for performance assessment and the first music dataset in the field of KT, with abundant data collected by a public intelligent sight-singing practice platform, "SingMaster." Unlike the existing KT datasets, each question in "SingPAD" is defined as a note in a music score, and learners' music performance can be evaluated objectively and automatically utilizing music information retrieval technology. Several classical knowledge tracing models are tested on "SingPAD," and the experimental results show that "SingPAD" exhibits good consistency and discriminability with existing datasets. "SingPAD" can be used as a benchmark dataset for applying knowledge tracing models to predict music knowledge mastery levels and promote the development of knowledge tracing research. [For the complete proceedings, see ED675485.] |
| Abstractor: | As Provided |
| Notes: | https://github.com/itec-hust/singKT-dataset |
| Entry Date: | 2025 |
| Accession Number: | ED675668 |
| Database: | ERIC |
| Abstract: | Knowledge tracing (KT) aims to model a learner's knowledge mastery level through his historical exercise records to predict future learning performance. Using this technology, learners can get appropriate customized exercises based on their current knowledge states, and thus the great potential of personalized teaching services such as intelligent tutoring systems and learning material recommendations can be stimulated. Currently, the mainstream datasets in KT include ASSISTments, EdNet, STATICS2011, etc., which are mainly based on objective testing data in the fields of mathematics and language, lacking of datasets on music performance assessment. Therefore, based on the context of performance assessment in music education, specifically sight-singing evaluation, we introduce "SingPAD," the first dataset for performance assessment and the first music dataset in the field of KT, with abundant data collected by a public intelligent sight-singing practice platform, "SingMaster." Unlike the existing KT datasets, each question in "SingPAD" is defined as a note in a music score, and learners' music performance can be evaluated objectively and automatically utilizing music information retrieval technology. Several classical knowledge tracing models are tested on "SingPAD," and the experimental results show that "SingPAD" exhibits good consistency and discriminability with existing datasets. "SingPAD" can be used as a benchmark dataset for applying knowledge tracing models to predict music knowledge mastery levels and promote the development of knowledge tracing research. [For the complete proceedings, see ED675485.] |
|---|
