Machine Learning, a type of artificial intelligence that apply statistical learning techniques to point out patterns in data automatically to make highly accurate predictions. (“Definition of Machine Learning,” n.d.).
Generally, the knowledge about the data usage for each node can be acquired through the machine learning based time invariant fragmentation method (MLTIF). It helps to schedule the allocation and selective update for a specific time period.
In the distributed database system (DDBS), a computer network is interconnecting the geographically dispersed databases, and the administration is done by a distributed database management system (DDBMS). Among the strategies that are adopted to allocate the data fragments in a DDBS, one data allocation strategy allows a single copy of the database to be stored in the network. , As a result, the overall system communication cost, enquiry response time, and other factors can be minimised.
This time invariance concept enables to detect data usage patterns from the query history autonomously of the given database, identify time-invariant and their own time windows, and distribute these fragments to the nodes such that data communication and modernise synchronizations costs are minimized.
For the time invariant fragment (TIF), the values of each component attribute in a TIF are constant throughout this time interval. TIF’s are constructed from the query history of the entire system. The large amount of data of the query are needed to acquire knowledge about the retrieval patterns from the query history suggests the use of a machine learning technique. One of the roles of machine learning is to acquire knowledge from available data and use it to create new theories about the domain in question, in a fully automated manner.
While algorithm for creation and allocation of TIF’s is used by MLTIF for each site based on the query histories of the entire system. Whereas the evaluation of the MLTIF approach, the performance will be compared with non-replication, full-replication, and materialized view approaches using simulation.
In conclusion, the creation and allocation of the time invariant fragments with machine learning approach can brought a lot of advantages to the working environment. It helps to schedule the allocation of data fragments in a distributed database environment.