In order to be able to perform multimedia searches (like sounds, videos, images, etc.) we have to use data structures like the Spatial Approximation Tree (SAT). This structure is a nice example of a tree structure in which well-known tricks for tree parallelization simply do not work. It is too sparse, unbalanced and its performance is too dependent on the work-load generated by the queries being solved by means of searching the tree. The complexity measure is given by the number of distances computed to retrieve those objects close enough to the query. In this paper we examine some alternatives to parallelize this structure through the MPI library and the BSPpub library.
Most information in science, engineering and business has been recorded in form of text. This information can be found online in the World-Wide-Web. One of the major tools to support information access are the search engines which usually use information retrieval techniques to rank Web pages based on a simple query and an index structure like the inverted lists. The retrieval models are the basis for the algorithms that score and rank the Web pages. The focus of this presentation is to show some inverted lists alternatives, based on buckets, for an information retrieval system. The main interest is how query performance is effected by the index organization on a cluster of PCs. The server design is effected on top of the parallel computing model Bulk Synchronous Parallel-BSP.
The Web has became an obiquitous resource for distributed computing making it relevant to investigate new ways of providing efficient access to services available at dedicated sites. Efficiency is an ever-increasing demand which can be only satisfied with the development of parallel algorithms which are efficient in practice. This tutorial paper focuses on the design, analysis and implementation of parallel algorithms and data structures for widely-used text database applications on the Web. In particular we describe parallel algorithms for inverted files and suffix arrays structures that are suitable for implementing search engines. Algorithmic design is effected on top of the BSP model of parallel computing. This model ensures portability across diverse parallel architectures ranging from clusters to super-computers.