New algorithms for block segregated multiple key search strategy

• Authors: Ahmed Tarek
• Venue: CIS'09 Proceedings of the international conference on Computational and information science 2009
• Year: 2009

In this paper, two new algorithms for list segregated multiple key search strategies are discussed. In block search, a given list of elements, which are in sorted order is subdivided into a number of equal sized blocks, and the given key element is compared to the upper bounding element in each block until a bounding element is found, which is either equal to or greater than the key. If the bounding element is greater, then the search is confined in the corresponding block to locate the correct key position. In the proposed algorithms, the block search effort is confined to identify multiple key elements in a given list. The search efforts are optimized after identifying each individual key position by discarding a part of the search space (a portion of the list elements). Here, the list of keys are assumed to be sorted in ascending order, though it is possible to consider a descending list of keys with an ascending list of elements. Once a key index position is identified, the successive keys are searched for at the same or at the higher level blocks containing larger elements. This strategy considerably reduces the search efforts with multiple keys with increasing number of keys in the list. To search for the current key element, the algorithm simply discards the blocks containing the previous smaller keys, and starts at the bounding element of the latest block holding the immediately preceding key. Computational complexity of the proposed algorithms are also considered. Another modification uses the binary search instead of the linear search to identify the correct key position within the block, since the list is in sorted order. With large block sizes, this strategy considerably improves the search time. Therefore, the performance issues of the proposed algorithms are also taken into consideration.