NRank: A Unified Platform Independent Approach for Top-K Algorithms

What is it about?

Due to increasing capacity of storage devices and speed of computer networks during last years, it is still more required to sort and search data effectively. A query result containing thousands of rows from a relational database is usually useless and unreadable. In that situation, users may prefer to define constraints and sorting priorities in the query, and see only several top rows from the result. This paper deals with top-k queries problems, extension of relational algebra by new operators and their implementation in a database system. It focuses on optimization of operations join and sort. The work also includes implementation and comparison of some algorithms in standalone .NET library NRank.

Featured Image