net.obsearch.index.ghs
Class CompressedBitSet64

java.lang.Object
  net.obsearch.index.ghs.CompressedBitSet64

public class CompressedBitSet64
extends Object

CompressedBitSet64 stores bits in a byte array. The bit set must be first created (stored in a temporary file) And then the bytes will be loaded into memory. The compressed bit set works on longs and it allows sequential k-nn searches of longs with the hamming distance. Insertions must be done in ascending order. The main assumption is that compression will allow small indexes that will be stored in memory.

Author:

Arnoldo Jose Muller Molina

Field Summary

protected int	count
protected byte[]	data
protected long	first

Constructor Summary

CompressedBitSet64()
Create a new compressed bit set.

Method Summary

void	add(long bit) Add the ith bit to this bitset.
int	bucketDistance(long a, long b) Hamming distance used for searching.
void	commit() We will stop adding values and now we will use the bit set for search.
protected long[]	getAll() Return all the buckets just for debugging purposes.
long	getBytesSize()
protected long	read(it.unimi.dsi.io.InputBitStream in)
long[]	searchBuckets(long query, int maxF, int m) Search the maxF closest buckets by hamming distance for the given query
List<OBResultInvertedByte<Long>>	searchFull(long query) Search the entire bitset.
int	size()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

data

protected byte[] data

count

protected int count

first

protected long first

Constructor Detail

CompressedBitSet64

public CompressedBitSet64()
                   throws OBException

Create a new compressed bit set.

Throws:

IOException

OBException

Method Detail

add

public void add(long bit)
         throws OBException

Add the ith bit to this bitset.

Parameters:

bit -

Throws:

OBException

commit

public void commit()
            throws OBException

We will stop adding values and now we will use the bit set for search.

Throws:

OBException

getBytesSize

public long getBytesSize()

bucketDistance

public final int bucketDistance(long a,
                                long b)

Hamming distance used for searching.

Parameters:

a -

b -

Returns:

searchBuckets

public long[] searchBuckets(long query,
                            int maxF,
                            int m)
                     throws OBException

Search the maxF closest buckets by hamming distance for the given query

Parameters:

query - The query we will use

maxF - The number of objects that will be searched.

m - the number of bits == max distance expected.

Returns:

The closest objects to the given query.

Throws:

InstantiationException

IllegalAccessException

OBException

searchFull

public List<OBResultInvertedByte<Long>> searchFull(long query)
                                            throws OBException

Search the entire bitset. This should be done with datasets that can fit in memory and only to generate statistics on the data.

Parameters:

query - query to search

Returns:

Throws:

OBException

size

public int size()

getAll

protected long[] getAll()
                 throws IOException

Return all the buckets just for debugging purposes.

Returns:

Throws:

IOException

read

protected long read(it.unimi.dsi.io.InputBitStream in)
             throws IOException

Throws:

IOException