In a product line engineering, several studies have been made on analysis of feature which determines commonality and variability of product. Fundamentally, because the studies are based on developer's intuition and domain expert's experience, stakeholders lack common understanding of feature and a feature analysis is informal and subjective. Moreover, the reusability of software products, which were developed, is insufficient. This paper proposes an approach to analyzing commonality and variability of features using semantic-based analysis criteria which is able to change feature model of specific domain to feature-ontology. For the purpose, first feature attributes were made, create a feature model following the Meta model, transform it into feature-ontology, and save it to Meta feature-ontology repository. Henceforth, when we construct a feature model of the same product line, commonality and variability of the features can be extracted, comparing it with Meta feature ontology through a semantic similarity analysis method, which is proposed. Furthermore, a tool for a semantic similarity-comparing algorithm was implemented and an experiment with an electronic approval system domain in order to show the efficiency of the approach Was conducted. A Meta feature model can definitely be created through this approach, to construct a high-quality feature model based on common understanding of a feature. The main contributions are a formulating a method of extracting commonality and variability from features using ontology based on semantic similarity mapping and a enhancement of reusability of feature model.