A reference-point-based multi-objective materialized view selection algorithm

Data warehouse plays a pivotal role in devising business strategies for organizations to stay competitive in the market. Analytical queries posed in the data warehouse must be answered efficiently. Using materialized views can be an effective strategy for reducing the response time of certain queries. Although materializing all potential views could enhance query performance, but this approach is impractical due to limited storage capacity. Moreover, choosing the optimal views is a problem of the NP-complete class. Consequently, a subset of views must be chosen to minimize query response time while adhering to storage constraints. This paper addresses this challenge by proposing a view selection algorithm that employs the reference-point-based non-dominated sorting algorithm (NSGA-III) to select views from a multi-dimensional lattice. The proposed algorithm is compared with the existing multi-objective view selection algorithms.