Package org.terrier.structures.indexing.classical

Class BlockInvertedIndexBuilder

  • public class BlockInvertedIndexBuilder
extends InvertedIndexBuilder
    Builds an inverted index saving term-block information. It optionally saves term-field information as well.

    Algorithm:

    1. While there are terms left:
      1. Read M term ids from lexicon, in lexicographical order
      2. Read the occurrences of these M terms into memory from the direct file
      3. Write the occurrences of these M terms to the inverted file
    2. Rewrite the lexicon, removing block frequencies, and adding inverted file offsets
    3. Write the collection statistics

    Lexicon term selection: There are two strategies of selecting the number of terms to read from the lexicon. The trade-off here is to read a small enough number of terms into memory such that the occurrences of all those terms from the direct file can fit in memory. On the other hand, the less terms that are read implies more iterations, which is I/O expensive, as the entire direct file has to be read for every iteration.
    The two strategies are:

    • Read a fixed number of terms on each iterations - this corresponds to the property invertedfile.processterms
    • Read a fixed number of occurrences (pointers) on each iteration. The number of pointers can be determined using the sum of frequencies of each term from the lexicon. This corresponds to the property invertedfile.processpointers.
    By default, the 2nd strategy is chosen, unless the invertedfile.processpointers has a zero value specified.

    Properties:

    • invertedfile.processterms - the number of terms to process in each iteration. Defaults to 25,000
    • invertedfile.processpointers - the number of pointers to process in each iteration. Defaults to 2,000,000, which specifies that invertedfile.processterms should be read from the lexicon, regardless of the number of pointers.
    Author:
    Douglas Johnson & Vassilis Plachouras & Craig Macdonald

    • Method Detail

      • getLexScanner

        protected org.terrier.structures.indexing.classical.InvertedIndexBuilder.LexiconScanner getLexScanner​(java.util.Iterator<java.util.Map.Entry<java.lang.String,​LexiconEntry>> lexStream,
                                                                                                      CollectionStatistics stats)
                                                                                               throws java.lang.Exception
        Overrides:
        getLexScanner in class InvertedIndexBuilder
        Throws:
        java.lang.Exception

      • createInvertedIndex

        public void createInvertedIndex()
        This method creates the block inverted index. The approach used is described briefly: for a group of M terms from the lexicon we build the inverted file and save it on disk. In this way, the number of times we need to read the direct file is related to the parameter M, and consequently to the size of the available memory.
        Overrides:
        createInvertedIndex in class InvertedIndexBuilder

      • traverseDirectFile

        protected void traverseDirectFile​(gnu.trove.TIntIntHashMap codesHashMap,
                                  gnu.trove.TIntArrayList[][] tmpStorage)
                           throws java.io.IOException
        Traverses the direct files recording all occurrences of terms noted in codesHashMap into tmpStorage.
        Overrides:
        traverseDirectFile in class InvertedIndexBuilder
        Parameters:
        codesHashMap - contains the term ids that are being processed in this iteration, as keys. Values are the corresponding index in tmpStorage that information about this terms should be placed.
        tmpStorage - Records the occurrences information. First dimension is for each term, as of the index given by codesHashMap; Second dimension contains fieldCount+4 TIntArrayLists : (document id, term frequency, field0, ... fieldCount-1 , block frequencies, block ids).
        Throws:
        java.io.IOException - if there is a problem while traversing the direct index.

      • writeInvertedFilePart

        protected long[] writeInvertedFilePart​(java.io.DataOutputStream dos,
                                       gnu.trove.TIntArrayList[][] tmpStorage,
                                       int _processTerms)
                                throws java.io.IOException
        Writes the section of the inverted file
        Overrides:
        writeInvertedFilePart in class InvertedIndexBuilder
        Parameters:
        dos - a temporary data structure that contains the offsets in the inverted index for each term.
        tmpStorage - Occurrences information, as described in traverseDirectFile(). This data is consumed by this method - once this method has been called, all the data in tmpStorage will be destroyed.
        _processTerms - The number of terms being processed in this iteration.
        Returns:
        the number of tokens processed in this iteration and the number of bytes of temporary mem that were used
        Throws:
        java.io.IOException

      • main

        public static void main​(java.lang.String[] args)
                 throws java.lang.Exception
        Use this main method to recover the creation of an inverted index, should it fail
        Throws:
        java.lang.Exception