Conformed Dimension Identification on Fusion Cubes Using Synonym, Hypernym, and Hyponym

The complex analysis needs of decision-makers may require variety data cubes that are spread over heterogeneous cubes. The decision-makers need to obtain as many relevant cubes as possible according to their queries. In this condition, the decision-makers need to combine heterogeneous cubes into new single cube (which is called fusion cubes process). This makes the conformed dimensions identification becomes necessary. Conformed dimensions are dimensions that represent the same objects in the real world, as links between cubes to be merged in fusion cubes. In previous studies, conformed dimensions identification in the fusion cubes was carried out using syntactic similarity with the Jaro-Winkler algorithm and semantic similarity with synonym relation between dimensions. However, not all conformed dimensions are identified. This affects the cubes that should be relevant are not included in the fusion cube. Therefore, this study tries to improve the conformed dimensions identification by adding hypernym and hyponym besides synonym in the conformed dimensions identification method. The proposed method presents a higher recall value than the method using only synonym. This shows that the use of hypernym and hyponym can improve the search for relevant cubes. Meanwhile, the proposed method results lower precision than the method using only synonym. This shows that the error rate of the proposed method is higher than the method using only synonym. However, based on F-measure, that is the balance score of recall and precision, the proposed method has a better F-measure value than method using only synonym.