Ontology Evaluation through Text Classification

A Novel Method for Text Classification Dealing with Local Data Structures and High Data Dimensionality

“…However, the added penalty term λ‖𝒎‖ 2 in objective ( 8) only decreases the weights of less relevant or irrelevant neighbors, but not selecting neighbors. To enable neighbor selection, a 𝑙 1 − 𝑛𝑜𝑟𝑚 regularizer is added to the objective to set the weights of those neighbors to be dropped to zero min𝑖𝑚𝑖𝑧𝑒 𝑓(𝒎) = ‖𝑩 𝑇 𝒎 − 𝒙 ̃‖2 + 𝜆 1 ‖𝒎‖ 2 + 𝜆 2 ‖𝒎‖, 𝐦 ∈ 𝐑 k , 𝜆 1 , 𝜆 2 ∈ 𝐑 + (10) By adding the 𝑙 2 − 𝑛𝑜𝑟𝑚 regularizer and the 𝑙 1 − 𝑛𝑜𝑟𝑚 regularizer, not only a unique correlation vector could be obtained for each test sample, but also the most relevant neighbors among the k nearest neighbors of a test sample will be selected. In the field of statistical learning, objective function (10) is called an elastic net regression.…”

“…To enable neighbor selection, a 𝑙 1 − 𝑛𝑜𝑟𝑚 regularizer is added to the objective to set the weights of those neighbors to be dropped to zero min𝑖𝑚𝑖𝑧𝑒 𝑓(𝒎) = ‖𝑩 𝑇 𝒎 − 𝒙 ̃‖2 + 𝜆 1 ‖𝒎‖ 2 + 𝜆 2 ‖𝒎‖, 𝐦 ∈ 𝐑 k , 𝜆 1 , 𝜆 2 ∈ 𝐑 + (10) By adding the 𝑙 2 − 𝑛𝑜𝑟𝑚 regularizer and the 𝑙 1 − 𝑛𝑜𝑟𝑚 regularizer, not only a unique correlation vector could be obtained for each test sample, but also the most relevant neighbors among the k nearest neighbors of a test sample will be selected. In the field of statistical learning, objective function (10) is called an elastic net regression. Since least square loss function is convex, 𝑙 1 − 𝑛𝑜𝑟𝑚 regularizer and 𝑙 2 − 𝑛𝑜𝑟𝑚 regularizer are convex, and the summation of convex functions are convex, objective function (10) is convex.…”

“…In the field of statistical learning, objective function (10) is called an elastic net regression. Since least square loss function is convex, 𝑙 1 − 𝑛𝑜𝑟𝑚 regularizer and 𝑙 2 − 𝑛𝑜𝑟𝑚 regularizer are convex, and the summation of convex functions are convex, objective function (10) is convex. However, it is not smooth at 𝒎 = 𝟎 because of the 𝑙 1 − 𝑛𝑜𝑟𝑚 regularizer.…”

“…Text categorization is the study of classifying text documents by assigning them to one of the pre-defined classes. Many researchers have put efforts on text categorization and their results have been used in various domains such as webpage classification [2][3][4], sentiment identification [5,6], spam filtering [7][8][9], and search result disambiguation [10,11].…”

Ontology Validation Algorithm on Data Driven Approach and Vocabulary Aspect

“…This approach is a hybrid of several other approaches from semantic similarity measurement and vocabulary approach. While other works state user profiling as their intermediate data driven approach [2], data driven to text classification [3], Noy et al [4] adds the markup of user based ontology selection history on search terms to draw the analysis of ontology selection from the BioPortal ontology repositories result.…”