Disequilibrium multi-dividing ontology learning algorithm

Approximation analysis of ontology learning algorithm in linear combination setting

“…We take half of vertices as ontology sample, i.e., |S| = 13. Similarly, to compare it with other ontology learning algorithms, we directly use the experimental data which were presented in the [24,25] and [27]. Furthermore, we test the accuracy of "confidence weighted ontology algorithm" presented in [37] and "weak function based ontology learning algorithm" manifested in [38], and compare to our ontology learning algorithm.…”

“…In contrast to other ontology learning algorithms compared in experiments, some of them are not designed for tree structures, and some of them use different angles to design algorithms. For instances, (1) although the confidence weighted ontology algorithm in [37] is also designed under a multi-dividing framework, its purpose is to save space complexity, and its core algorithm is a buffer update strategy, not an iteration of ontology functions; (2) disequilibrium ontology learning in [27] is also presented in multi-dividing ontology learning setting, while it focuses on the balance between the data rather than the structure of the ontology graph. In general, the efficiency of the algorithm in this paper reflects its advantages over tree-structured ontology graphs.…”

“…With the proliferation of ontology processing concepts, machine learning algorithms are applied to ontology similarity calculations (some ontology learning tricks can be referred to Gao et al [21][22][23] and [24]). Among them, the ontology learning algorithm in multi-dividing setting has proved to be more efficient for the similarity calculation under the tree-shaped ontology structures (see Gao et al [25], Gao and Farahani [26], Wu et al [27], Sangaiah et al [28] and Gao and Xu [29] for more details). Due to the engineering accuracy of the multi-dividing ontology learning algorithm has confirmed by different ontology applications, this paper no longer gives the experimental results of the algorithm under special ontology data, but from the statistical point of view.…”