A novel automatic property weight generator for semantic data integration

The dramatic increase of heterogeneous data resources, whether they are semantic knowledge bases or databases, demands for an automatic data integration technique, which is directly affected by the weight of property associated to instances or data. Choosing a suitable metric for generating weight for a property automatically is nevertheless a formidable task. In this study, we analyze different metrics for generating weights and formulate mathematical reasoning to select an appropriate one. Our observation suggests that the weight of a property is highly influenced by the number of instances contain the property, the number of instances contain the distinct value for the property and the total number of instances in a training dataset. The experiments and evaluations illustrate the fact and shows the strength of automatic weight generator for properties in a data integration technique.