Fuzzy Ontology-based Semantic Retrieval Over Relational Databases

The data stored in relational databases has fuzzy semantics. For example, "high", "short", "fat", "thin", "moderate", "normal" are fuzzy data. The NULL value is the incomplete information stored in relational databases, which also leads to the uncertainty of relational database. Fuzzy ontology is to describe the shared conceptual model of fuzzy concepts and communication between them, which can express the fuzzy semantics of natural language. The fuzzy ontology can greatly improve the effect of semantic retrieval. Therefore, in this paper fuzzy ontology is employed to express fuzzy semantics in relational databases and fuzzy ontology-based semantic retrieval technologies in relational databases arel studied to improve the effect of semantic retrieval over relational databases.Relational database normalization design leads to complete entity (object) information is decomposed to be stored in many pluralistic groups, making the relational data stored by relation mode in relational database has incomplete semantic. Also the metadata itself defined by relational database model designing has fuzziness.The fuzzy semantics in relational databases is not favorable for relational database information retrieval. Based on relational data to semantic data mapping language (such as R2RML language) relational data is mapped to RDF semantic data, to express the semantics of the structured data. In this paper the existing domain ontology is used to construct fuzzy ontology, to express fuzzy semantics in relational database, and a fuzzy ontology-based semantic retrieval model is put forward the semantic retrieval algorithm and the sorting algorithm are designed and implemented, and a prototype system is implemented. The system can realize the fuzzy semantic retrieval over relational databases, and can effectively retrieve the data with fuzzy semantics in the relational databases. Moreover, the recall and precision of semantic retrieval over relational databases can be effectively improved.To validate the effectiveness of the fuzzy ontology-based semantic retrieval model proposed in this paper, DBLP database is used as test data sets, and fuzzy ontology based on domain ontology is constructed. With the fuzzy ontology, the relational data is converted to RDF data by using semantic mapping language R2RML. At last, we do the semantic retrieval based on the RDF data. The analysis of the experimental results shows that the proposed semantic retrieval system outperforms the traditional semantic retrieval system in terms of recall and precision.