Design Of Ontology-based Semantic Retrieval Engine For Animation Material And Ranking Research

With the economic globalization and commercial integration, animation industry flourishes.Because manufacturing animation is time-consuming,laborious and low productive,managing and effective reusing the large number of animation materials at hand make sense.It is difficult for traditional full-text retrieval to express users' intent of retrieval.Moreover,it is only a literal meaning match,rather than semantic meaning.The text-based or content-based retrieval model is not easy to resolve the "semantic gap" between the users and the computers. Therefore,semantic-based retrieval emerges.This paper is mainly about the studies of ontology-based information retrieval.Semantic Retrieval Subsystem as well as ranking algorithm of the system is designed and realized with the use of ontology technology and inverted index mechanism of Lucene.Firstly,related research on image retrieval is introduced and the advantages and disadvantages of the retrieval are analyzed.Then the related theory of ontology is introduced, and the function of the ontology in the information retrieval system and the ontology-based model of information retrieval are also analyzed. Additionally,the semantic model of animation materials is put forward. Finally,the architecture and the realization way of Ontology-based Semantic Retrieval Engine for Animation Matieral are proposed.To narrow "semantic gap" in the process of the retrieval,the extension algorithm of concepts based on ontology is brought forward. Furthermore,the algorithm named weighted calculation method of concepts similarity based on distance and density is advanced according to the hierarchical relationship between concepts in ontology,which is used to measure the similarity between initial query terms and expanded terms.For the sake of reflecting the users' intentions,the genetic algorithm for re-combination and optimization of query terms is designed. Through analyzing the ranking algorithm of search engine,and adopting the ranking model of traditional information retrieval and the default sort mechanism of Lucene,a new sorting algorithm is put forward.。In order to verify the feasibility of the system design and validity of algorithms in this paper,Ontology-based Semantic Retrieval Engine for Animation Material has been implemented by using of rapid prototyping method.The experimental results show that the recall and precision in the system are improved to a certain extent.