Research On Semantic Retrieval And Scenario Generation Key Technologies Based On Animation Domain

Animation generation technology is a new process of animation generation which integrates artificial intelligence and multimedia technology. The system accepts story scripts described in natural language, then analyzes and comprehends the story, distill the information of scenario, event, actors and so on by using related technologies such as artificial intelligent, semantic retrieval, spatial reasoning and layout technologies, finally generates animation by using semantic annotation to make the multimedia information (cartoon resources) semantic change and using semantic retrieval engine to inquire and reason related scripts with assistance of the knowledge base.This paper focuses on the key technologies of the semantic retrieval and Scenario generation in the field of of animation. In the first half of the paper, we mainly study the key technologies of Semantic retrieval in the field of animation. By analyzing some of the key technologies in the semantic search, and propses ontology-based query expansion mechanism and the weighted sort optimization algorithm, combined with the specific characteristics of Ontology, and presents query-oriented ontology reasoning architecture. To with LarKC development platform, and preliminary implements a query-oriented animation semantic retrieval system by using SPAQRL in LarkC. This system uses the OWL as Ontology description language, and establishs Ontology knowledge base and rule base based on the field of animation, and uses semantic annotation to the semantics of animation material, with Jena to implement ontology reasoning and query based on animation domain knowledge, and tests query efficiency, recall and precision ratio by a series of experiment. The experiment results have proved that the system can meet the retrieval requirements, and has higher recall and precision. It verifies the practicability and validity of the reasoning framework. The latter part of the paper, we mainly study the key technologies of Scenario generation, and discuss the scenario generation from both qualitative and quantitative. By analyzing some of the key technologies and methods in the Scenario generation, and propses a scenario generation algorithm and scene object space relationship layout reasoning method to used to calculate the position and orientation of objects in the scene and so on. Finally, we propose a framework and design ideas of the scenario generation system. For a story script, animation ontology Language can describe it in form of RDF/N3. In other words, we can use RDF triples form to describe a story script, or express a section of script with the form of RDF file. Then inquire the action which match the animation ontology language with SPARQL query language, we divide the story to scenarios by means of "Action".Then a typical scenario description is generated by a pattern in scenario database which matches its background. At last, the qualitative data would be dynamically modified by analyzing of time and spatial information in animation ontology language and object rule database. Finally, we add the figure in the scenario and get the final scenario.