| The problem of uncorrelated or irrelated keys to the retrieval term in traditionalinformation retrieval leads to the difficulties of querying for users from various fields,and it also greatly reduces the recall and precision ratio. Semantic search has beensuggested to improve the results. By the method of building semantic relationshipsbetween resources from the point of knowledge comprehension, it not only solves thematter of irrelevant term to the retrieval term to some extent, but also improves thesearching effect. While the results still have difference in different context, andsometimes even don’t fit for the professional knowledge. Ontology as the basis ofknowledge exchanging, describes and provides the understanding of konwlege. Thestudy of ontology based semantic relevant discovery will be great value in retrieval.The design of semantic relevant discovery model in this paper makes greatprogress in retrieval, the results which are correlated to the retrieval term, well meetwith the needs of users, and accords with the domain knowledge. By parsing theconstructed domain ontology, we calculate the correlations from the semanticdistance of depth and breadth in ontology, and the results are collected together toimprove and perfect the Semantic Knowledge Base. Compared with the traditionalmethods, the experimental data in agricultural sector shows that the novel modelbetter corresponded with domain reality. It not only gets the degree of association ofthe defined related terms according as the structure of ontology, but aslo calculatesthe degree of association of the indirect terms by the reference term.We design an ontology based tea semantic retrieval system on the basis of themodel. The system will output related terms sorted by their degrees of association tothe retrieval terms. The realization of tea semantic retrieval system verifies thefeasibility and practicability of the relevant discovery model, accordingly, it is reallyworthwhile to other fields. |
PDF Full Text Request