Construction And Research Of Metal Material Picture Knowledge Base Based On Wikipedia

At present,the image resource becoming more and more abundant,the difficulty of finding the needed image resource gradually increases.The traditional image library cannot reflect the semantic relations among images,but the image knowledge base can show the relevant pictures that meet the user’s requirements according to the semantic information of user’s different choices.Hence,in this paper,a construction strategy of the image ontology model is proposed based on the metallic materials domain data and an approach is designed to extract metallic materials images data from Wikipedia to build picture knowledge base,in order to facilitate the user’s multi-facet semantic query.1)Metallic materials image data is obtained from Wikipedia.All hyperlinked pages are obtained by taking Metal page in Wikipedia as the access to find pages related to metallic materials using Naive Bayes classification algorithm and the pictures of metallic materials from these pages are achieved.2)The image metadata acquisition and analysis strategies are designed based on the Wikipedia.First,a metadata model for metallic materials picture is constructed by analyzing the content of the web page related to the metallic materials,and the picture metadata is obtained based on the model.Then,the picture classification system is determined by the metadata.and the image categories are determined by using the Naive bayes algorithm and SVM algorithm to classify the pictures,Finally,form the metallic materials pictures knowledge base by using the similarity algorithm that is to calculate the semantic relevance of the picture.3)The metallic materials picture knowledge base is linked to the existing ontology.First,the link properties are determined,which build the relationships between the metallic materials picture knowledge base and existing metallic materials ontology STSM.Then,by using string comparison algorithm to calculate the similarity between the properties,the ontology linking is realized.In addition,the approaches of obtaining the web pages related to metallic materials,image classification algorithm and similarity calculation algorithm are evaluated in light of F1-measure and time performance.The experimental results demonstrate that the proposed approaches are feasible,and the time performance is acceptable.Meanwhile,a prototype system is designed to visually display the metallic materials picture knowledge base,and the process of acquiring picture data is shown in detail.