Research On The Ontology-based Sematic Search In Mass Agriculture Information

How to process and retrieval massive, distributed and heterogeneous informationhas become a key issue, which need to be resolved during the development andapplication of information technology. With the explosive growth of information andthe changeable requirements, current information retrieval methods shows somelimitations, such as the machine can’t understand and process the accurate semanticinformation, and the precision/recall of retrieval results still low. Ontology is aconceptual modeling for specific domain in the real world, which performs better indescribing the semantics of domain knowledge. It is an effective way to solve theabove problems by introducing domain ontology into information retrieval. Therefore,the study of ontology-based sematic search in mass agriculture information canimprove the precision/recall of retrieval results and the quality of the agricultureinformation retrieval.To improve the precision/recall of semantic retrieval results, this paper optimizeagriculture ontology, corpus and semantic retrieval technology.(1) Propose a clumping method based on mappings for fully mining, sharing andreusing of ontologies. In the process of ontology mapping, we clump the sourceontology and target ontology respectively for building high quality mappings, andthen dig out the potential relationships among entities for improving the quality ofdomain ontology.(2) Propose a domain-oriented noise removal algorithm to produce better corpus.We collect the mass agriculture information from internet by using web crawler tool,and then extract and organize the collected information for building corpus.(3) Introduce the application of association rules, and improve the classicassociation rules mining algorithm “Apriori” for optimizing semantic retrievaltechnology. On the basis of information form domain ontology and corpus, we minithe potential rules by using improved mining method, and then improve theprecision/recall of retrieval system with the rules.We design and implement the sematic search system for mass agriculture information retrieval. This system translates the query into semantic queryautomatically after users enter keywords, and calculates semantic similarity with thedata in the current repository. Finally, according to the semantic similarity values, itsorts the results by descending order, and then presents to the user by list. The resultsprove the feasibility and efficiency of this work, and provide a valuable reference forresearch on semantic retrieval technology in other domains.