Knowledge Graph Entity Similarity Calculation under Active Learning

• Published in: Complexity (New York, N.Y.) Vol. 2021; no. 1
• Main Authors: Li, Lianhuan, Zhang, Zheng, Zhang, Shaoda
• Format: Journal Article
• Language: English
• Published: Hoboken Hindawi 2021; Hindawi Limited; Wiley
• Subjects: Active learning; Algorithms; Alignment; Collaboration; Decomposition; Deep learning; Design; Distributed processing; Efficiency; Interactive systems; Internet; Knowledge representation; Machine learning; Methods; Principal components analysis; Questions; Recommender systems; Similarity; Sparsity
• DOI: 10.1155/2021/3522609
• ISSN: 1076-2787

To address the objectives of the adaptive learning platform, the requirements of the system in terms of business, functionality, and performance are mainly analysed, and the design of functions and database is completed; then, an updatable learner model is constructed based on the cognitive diagnosis model and resource preference attributes; then, the construction of the knowledge map is completed based on embedding to achieve knowledge point alignment, and based on this, the target knowledge points of learners are located with the help of deep learning; at the same time, the target knowledge points are taken as the starting point to generate the best learning path by traversing the knowledge map, and the corresponding learning resources and test questions are recommended for them with the help of the architecture; finally, the adaptive learning platform is developed in the environment using the architecture. Also, the target knowledge point is used as the starting point to traverse the knowledge map to generate the best learning path, and the corresponding learning resources and test questions are recommended for the learner in combination with the learner model; finally, this study adopts an architecture for the development of an adaptive learning platform in the environment to realize online tests, score analysis, resource recommendation, and other functions. A knowledge graph fusion system supporting interactive facilitation between entity alignment and attribute alignment is implemented. Under a unified conceptual layer, this system can combine entity alignment and attribute alignment to promote each other and truly achieve the final fusion of the two graphs. Our experimental results on real datasets show that the entity alignment algorithm proposed in this paper has a great improvement in accuracy compared with the previous mainstream alignment algorithms. Also, the attribute alignment algorithm proposed in this paper, which calculates the similarity based on associated entities, outperforms the traditional methods in terms of accuracy and recall.