org.ranksys.compression.codecs

Interface CODEC<T>

Type Parameters:

T - type of the output of the compression

All Superinterfaces:

Serializable

All Known Implementing Classes:

AbstractCODEC, BitStreamCODEC, EliasFanoBitStreamCODEC, FastPFORVBCODEC, FixedLengthBitStreamCODEC, FORVBCODEC, GammaBitStreamCODEC, IntegratedFORVBCODEC, IntegratedVByteCODEC, LemireCODEC, NewPFDVBCODEC, NullCODEC, OptPFDVBCODEC, RiceBitStreamCODEC, Simple16CODEC, VByteCODEC, ZetaBitStreamCODEC

public interface CODEC<T>
extends Serializable

Integer compression codec. Compression of arrays of integers into more compact representations of them.
If you use this code, please cite the following papers:

• Vargas, S., Macdonald, C., Ounis, I. (2015). Analysing Compression Techniques for In-Memory Collaborative Filtering. In Poster Proceedings of the 9th ACM Conference on Recommender Systems. http://ceur-ws.org/Vol-1441/recsys2015_poster2.pdf.
• Catena, M., Macdonald, C., Ounis, I. (2014). On Inverted Index Compression for Search Engine Efficiency. In ECIR (pp. 359–371). doi:10.1007/978-3-319-06028-6_30

The code that reproduces the results of the RecSys 2015 poster by Vargas et al. in a separated project: http://github.com/saulvargas/recsys2015
The search index compression technologies of the ECIR paper by Catena et al. is part of the Terrier IR Platform: http://terrier.org/docs/v4.0/compression.html.

Author:

Saúl Vargas (Saul.Vargas@glasgow.ac.uk)

Method Summary

Modifier and Type	Method and Description
T	co(int[] in, int offset, int len) Compress.
int	dec(T t, int[] out, int outOffset, int len) Decompress.
boolean	isIntegrated() Returns if the CODEC is integrated, i.e.
void	reset() Resets the collection of statistics.
long[]	stats() Returns statistics of the CODEC in terms of total input/output bytes processed.

Method Detail

co

T co(int[] in,
     int offset,
     int len)

Compress.

Parameters:

in - array to be compressed.

offset - offset of in

len - number of integers to be compressed

Returns:

compressed representation of the array

dec

int dec(T t,
        int[] out,
        int outOffset,
        int len)

Decompress.

Parameters:

t - compressed representation of an array.

out - array where to write the compressed.

outOffset - offset of out

len - number of integers that need to be decompressed

Returns:

currently not used

stats

long[] stats()

Returns statistics of the CODEC in terms of total input/output bytes processed.

Returns:

statistics of input/output bytes.

reset

void reset()

Resets the collection of statistics.

isIntegrated

boolean isIntegrated()

Returns if the CODEC is integrated, i.e. requires that the input is a list of sorted integers (no use of d-gaps required).

Returns:

true is the CODEC is integrated, false otherwise