Authors: Anjana Gosain; Kriti Saroha
Addresses: University School of Information, Communication and Technology, Guru Gobind Singh Indraprastha University, Dwarka, India ' School of Information and Technology, Centre for Development of Advanced Computing, Noida, India
Abstract: Bitemporal data warehouse (BTDW) design that extends the multidimensional model uses bitemporal (combination of valid time and transaction time) timestamps on the members of multidimensional data in order to manage time-varying data. Various approaches to manage changes of schema and dimension data in BTDW exist that use the combination of both valid-time as well as transaction time. Also, measures/facts can undergo change with time however current BTDWs do not handle or track changes for dynamic measures. This would pose problem for applications like fraud detection, where it is required to know and record the time when changes have occurred in the system; schema, dimension data or measures. Thus, the changes occurring in the measures need to be recorded indicating the time when a specific measure value is valid or updated. In this work, we propose solution for handling time-varying measures using bi-temporal versioning of fact table schema and fact data along with storage options for handling and storing the separate versions of fact table schema and data.
Keywords: data warehouse; bitemporal data warehouse; BTDW; schema versioning; transaction time; valid-time; bitemporal; facts; measures.
International Journal of Information Systems and Management, 2020 Vol.2 No.2, pp.93 - 105
Accepted: 05 May 2019
Published online: 23 Oct 2020 *