Research On Ship Domain Knowledge Acquisition Based On Ontology

Knowledge acquisition is one of the three key technologies in knowledge base construction, it is also a bottleneck problem of knowledge base construction, and the technology affects the popularization and application of knowledge base directly. Most of the traditional knowledge acquisition was completed manually, and the efficiency is so low that it is unable to meet the requirements for tremendous information processing currently. Therefore, automatic or semi-automatic knowledge acquisition is a hot spot on researching in knowledge engineering field at present. Furthermore, how to represent the knowledge so that we can improve the efficiency in the use of knowledge is another key problem in the process of knowledge base construction. Ontology knowledge representation is the most popular method of knowledge representation currently, in a word, it represents knowledge as ontology form, which improves the definition of knowledge and decreases the analysis cost in knowledge acquisition.After we have researched the existing knowledge acquisition method, the paper combines ontology theory with knowledge acquisition theory in ship field, and presents a framework in knowledge acquisition for ship domain based on ontology. The framework includes four basic modules:ship seed ontology introduction, text preprocessing, ship knowledge acquisition and ship ontology perfection. The task of ship seed ontology introduction module is that extracts core concepts and relationships of ship domain from "Chinese Thesaurus of Communication" and "Classified Chinese Thesaurus", and then constructs ship seed ontology. Test preprocessing module is mainly for parsing Chinese web pages on ship domain grasped from internet, and processing of Chinese word segmentation, etc, and then generates concept set. The ship knowledge acquisition module is to complete seed ontology parsing, concept matching and ontology editing. The paper introduces semantic similarity algorithm based on HowNet in the sub module of concept matching, and improves the sememe similarity algorithm. The task of ship ontology perfection module is to reason and evaluate to the expanded seed ontology, so that the knowledge of ship ontology will be more scientific and reasonable. At last, in this paper, we designed and implemented the knowledge acquisition prototype system for ship domain based on ontology, and verified its technical feasibility by experiments. This paper represents the ship domain knowledge as ontology form and implements semi-automatic knowledge acquisition, which increases the semantic feature of knowledge, improves the automatic degree of knowledge acquisition, speeds up the process of domain knowledge acquisition, resolves the problem of over-reliance on experts in the process of traditional knowledge acquisition, and also provides an alternative and effective method to domain ontology knowledge acquisition.