A Study Of Ontology Knowledge Base Construction In Rice Field

With the research of Ontology knowledge base raised, people are constructing Ontology’s domain gradually extended to many areas. However, due to Ontology own highly specialized, domain characterized and need relative long time to construct, the construction work of Ontology become a high intellectual labor-intensive work. The construction process of Ontology need a high level knowledge background of the specific domain, and the construction process is difficult to be automated. The most important, construction and maintenance of Ontology instance are heavily dependent on human intelligence work, so it is difficult to ensure the quality and the efficiency for instances extraction and construction of Ontology. After long time research of Ontology construction methods, in this paper, it uses Ontology reuse method to finish the concept of rice Ontology construction, designed to build the framework of Ontology instances, and implemented the rice Ontology instance semi-automatic extraction which based on neural network of deep learning methods. It provides a useful try on Ontology instance extraction, and shows new ideas and ways to semi-automated Ontology construction. this paper presents a naming entity extraction method based neural network deep learning system, which can make the build of Ontology instances more effective.The studies are as follows:1. Research on Ontology knowledge base construction, including the concept and instance of Ontology knowledge base.2. Named entity extraction method research, especially in big data environment, which means massive data, content complex and diverse environment, highly specialized knowledge-language environment, to study deeply in entity extraction method based on neural network algorithm, and validate the method operability and effectiveness compare with empirical methods.3. Proposed the neural network Ontology instance extraction method. Using entity extraction method based on the deep learning named entity name to achieve rice Ontology instance extraction, and compare with the experts’ work result in order to improve the empirical research. By collecting abstracts of rice Ontology Science and Technology in English literature, constructing the corpus, using machine learning method based on neural network semi-automatic extraction of rice cultivar name, combined with artificial identification and verification in order to accomplish rice Ontology construction instances(including 2326 concepts, 109 relations, 23 attributes and 537 instances).