Study On Ontology Learning And Its Application To Semantic Retrieval

Ontology as the formal specification of a shared conceptualization has good concept structure and semantic representation capabilities, it has drawn more and more attention from specialists and scholars both at home and abroad and has been widely used in Knowledge engineering, semantic retrieval etc. However, at this stage the construction of Ontology which is the basic task in Ontology research field mostly still uses manual method, it takes considerable human resources and time and even needs experts'participation, and it gradually becomes the obstacles of Ontology development and promotion. Therefore, it has become a significant research direction to construct Ontology through Ontology learning.This paper discusses and researches the methods of Ontology learning. Firstly, it briefly introduces the basic knowledge of Ontology and Ontology learning. Then, it introduces several typical methods of concept extraction in Ontology learning and analyzes the advantages and disadvantages of each method. On the basis of the method for concept extraction based on Bootstrapping, it increases the compound words extraction and improves the statistics method of frequency, which can be more scientific to extract the domain concept. At the meanwhile, aiming at the weakness which the concept extraction method based on Bootstrapping uses statistical methods and ignores the semantic affect to the result, this paper uses semantic similarity to measure the relevance between concepts, the accuracy of concept extraction is improved.In addition, this paper explores the relationship extraction in Ontology learning, it uses concept hierarchy clustering method to obtain the taxonomic relation. For the extraction of non-taxonomic relation, this paper proposes a method combined association rule and dependency parsing, which acquires the relationship between two concepts through the association rules and then uses dependency parsing to determine the semantic tag for the relationship.Finally, this paper implements an Ontology-based semantic retrieval system, the Ontology is constructed by the Ontology learning method this paper introduces, the effect of system application demonstrate that this Ontology learning method is very accurate and efficient.