A retrieval method of learners’ behavior features based on K-means clustering algorithm

• Published in: Cluster computing Vol. 27; no. 2; pp. 2049 - 2058
• Main Authors: Wang, Shaohua, Xu, Xiaoxiong
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.04.2024; Springer Nature B.V
• Subjects: Accuracy; Algorithms; Batch processing; Classification; Cluster analysis; Clustering; Coding; Coding standards; Computer Communication Networks; Computer Science; Disease; Efficiency; Hospitals; Information processing; Libraries; Library technology; Medical coding; Medical diagnosis; Medical records; Methods; Missing data; Neural networks; Operating Systems; Processor Architectures; Retrieval; Signal to noise ratio; Support vector machines; User behavior; Vector quantization; Missing data filling; Learner behavior; Feature retrieval; K-means clustering; Missing forest algorithm; Coding library
• ISSN: 1386-7857; 1573-7543
• DOI: 10.1007/s10586-023-04077-9

This paper studies the retrieval method of learners’ behavior features in the coding library the K-means clustering algorithm, which can effectively retrieve learners’ behavior features and ensure the safety and stability of the coding library. After constructing the medical coding database, using the missing forest algorithm to fill the missing data in the learner behavior data of the coding database, the improved binary K-means clustering algorithm is used, without setting the number of clusters, it is only necessary to carry out binary clustering operation on the behavior data of learners in the coding library after missing and filling, so as to obtain the behavior characteristics of learners in the coding library; These features are input as a support vector machine classifier. Through the classification training of support vector machine, the learner behavior features in the coding library can be classified and the corresponding retrieval results can be output. The experimental results show that this method can effectively retrieve learners’ behavior features in the code library and identify abnormal behaviors. The retrieval accuracy and efficiency are high, and it is less affected by the signal-to-noise ratio and the amount of data. It has significant advantages in the actual learners’ behavior features retrieval in the code library.