Efficiency

  • English
  • Italiano

Modern data-storage systems, such as Hadoop and BigTable use compression to reduce the volume of data to be transferred to/from disks, which are slow, as well as to increase the data to be stuffed into the internal memory of computers, which are fast. We have state-of-the-art results in this context about novel Lempel-Ziv schemes which improve Snappy and LZ4-solutions, by offering controlled trade-offs between compression ration and decompression speed. These results impact surprisingly on energy-efficiency issues of mobile/tablets and in the time-efficiency of streaming computations.

Another important topic in this area is data indexing of text/raw collections. We were the first  to  show that it is possible to design a compressed data structure which achieves entropy-bounded space and efficient query time for a large set of (sophisticated) queries. Nowadays we can safely admit that we know how to index compressed versions of structured (xml, graphs) and unstructured (text) data; this technology is mature to be spread into IR-applications.