Article Preview
TopIntroduction
A data warehouse is a combination of large volume of data useful for reporting, querying and analyzing in order to support decision makers. The design of DW architecture is fully dependent on data sources as well as user requirements (Bimonte et al. 2020). According to Inmon (2005), the data in DW is subject-based, consistent, integrated, time-dependent, non-volatile and suitable for decision support management. The main objective of DWs compared to databases is the incorporation of transactional data from various resources to make them accessible for analytical purpose. However, conceptual design in DW is a critical phase to correctly demonstrate the objectives and it is the principal component in which the perspective of decision makers and informatics must approve (Ballard et al., 1998, Prakash, & Prakash, 2019). As a consequence, it is more appropriate for designers to create a robust and consolidated conceptual model since the making is too expensive, and different software combining all processes of DWs providing pre-packaged solutions for any critical situations are recently available. Furthermore, the decision makers mostly rely on data about data called as metadata that is essential to analyse, model, build, utilize and interpret DW contents in improving the quality of decision (Gautam, 2018). Generally, a DW consists of fact table as well as dimension table where the former contains metrics or business facts located at the centre of schema surrounded by dimension tables. The fact table has certain columns for fact measures, foreign keys to access dimension table, and a primary key which is the combination of its entire foreign keys. Additionally, the fact table is responsible for storing the content of different measures of DW system.
From the last two decades, two methods namely normalized modelling (Inmon, 1991) and dimensional modelling (Kimball, 1996) are governing the implementation of DW in almost all sectors. Of these, dimensional modelling has been one of the predominant techniques that take DW implementation to next stage. Nevertheless, traditional dimension modelling does not support a new variable called temporal dimension which is popular nowadays. Furthermore, the data source of a traditional DW does not provide complete support for temporal data which contains several time-dependent aspects including different time stamps. This is due to the fact that traditional methods keep only the record of any one state of real world environment at a valid time (i.e. ability to view only one data at a time). However, the world as well as data formation changes with time thus there is a need for temporal database to keep track of data changes within a database environment (Ongoma, 2014). Thus, researchers focussed on the introduction of temporal dimension in DW architecture (Hultgren, 2012, Rönnbäck et al., 2010, Golec et al., 2017). Temporal data models can be developed by extending the conceptual model of traditional DW (like UML, ER and ORM) with temporal paradigms. Temporal dimension is a simple and new phenomenon that records all time-dependent variables, maintains history of databases and keeps track of variations so as to plan for the future (Gosain & Saroha, 2019). Besides, slowly varying dimensions in certain applications have been represented as temporal variables in the fact table as well as dimension table of DW framework (Araque, 2003, Phungtua-Eng & Chittayasothorn, 2019). Therefore, the knowledge acquired from both data warehouse and temporal dimensional modelling give rise to a new research area termed as temporal data warehouse (TDW).