A MknnIndex represents the index structure. More...

#include "../metricknn_c.h"

Functions
MknnIndex *	mknn_index_newPredefined (MknnIndexParams parameters, bool free_parameters_on_index_release, MknnDataset search_dataset, bool free_search_dataset_on_index_release, MknnDistance *distance, bool free_distance_on_index_release)
	Instantiates a new index identified by the given parameters. More...

MknnIndexParams *	mknn_index_getParameters (MknnIndex *index)
	Returns the parameters used to create or load the index. More...

MknnDataset *	mknn_index_getSearchDataset (MknnIndex *index)
	Returns the dataset used to create or load the index. More...

MknnDistance *	mknn_index_getDistance (MknnIndex *index)
	Returns the distance used to create or load the index. More...

const char *	mknn_index_getIdPredefinedIndex (MknnIndex *index)
	returns the id of the predefined index. More...

void	mknn_index_save (MknnIndex index, const char filename_write)
	Saves the built index to a file. More...

MknnIndex *	mknn_index_restore (const char filename_read, MknnDataset search_dataset, bool free_search_dataset_on_index_release, MknnDistance distance, bool free_distance_on_index_release, MknnIndexParams more_parameters, bool free_parameters_on_index_release)
	Loads an index from a file. More...

MknnResolver *	mknn_index_newResolver (MknnIndex index, MknnResolverParams parameters_resolver, bool free_parameters_on_resolver_release)
	Configures a new similarity search using the given index. More...

void	mknn_index_release (MknnIndex *index)
	Releases the index. More...

Detailed Description

A MknnIndex represents the index structure.

In order to create an index, at least two parameters must be provided:

a search dataset (MknnDataset), where the nearest neighbors will be searched in.
a distance (MknnDistance), the dissimilarity measure to be used to compare objects.

MetricKnn provides a set of pre-defined indexes identified by a unique distance identifier. Some valid IDs are:

LINEARSCAN The naive sequential scan of the search dataset.
LAESA The LAESA metric access method. http://dx.doi.org/10.1016/0167-8655(94)90095-7
FLANN-KDTREE The kd-tree, implemented by FLANN library.
FLANN-KMEANS The k-means tree, implemented by FLANN library.

The multidimensional indexes are implemented by the FLANN library. http://www.cs.ubc.ca/research/flann/

In order to simplify the creation of distances, there are function that returns the parameters for building the different indexes with mknn_index_newPredefined.

Note: The complete list of pre-defined indexes provided by MetricKnn can be printed by calling mknn_predefIndex_helpListIndexes. The parameters supported by each index can be printed by calling mknn_predefIndex_helpPrintIndex.

Function Documentation

MknnDistance* mknn_index_getDistance ( MknnIndex * index )

Returns the distance used to create or load the index.

Parameters

index the index

Returns: the distance used to create or load the index.

const char* mknn_index_getIdPredefinedIndex ( MknnIndex * index )

returns the id of the predefined index.

Parameters

index the index

Returns: the id of the predefined index.

MknnIndexParams* mknn_index_getParameters ( MknnIndex * index )

Returns the parameters used to create or load the index.

Parameters

index the index

Returns: the parameters used to create or load the index.

MknnDataset* mknn_index_getSearchDataset ( MknnIndex * index )

Returns the dataset used to create or load the index.

Parameters

index the index

Returns: the dataset used to create or load the index.

MknnIndex* mknn_index_newPredefined	(	MknnIndexParams *	parameters,
		bool	free_parameters_on_index_release,
		MknnDataset *	search_dataset,
		bool	free_search_dataset_on_index_release,
		MknnDistance *	distance,
		bool	free_distance_on_index_release
	)

Instantiates a new index identified by the given parameters.

Parameters

parameters	the definition of the index.
free_parameters_on_index_release	binds the lifetime of `parameters` to the new index.
search_dataset	the data to be indexed. Any search on this index will return some object from `search_data`.
free_search_dataset_on_index_release	binds the lifetime of `search_dataset` to the new index.
distance	a distance.
free_distance_on_index_release	binds the lifetime of `distance` to the new index.

Returns: a index (must be released)

MknnResolver* mknn_index_newResolver	(	MknnIndex *	index,
		MknnResolverParams *	parameters_resolver,
		bool	free_parameters_on_resolver_release
	)

Configures a new similarity search using the given index.

Parameters

index	the index that will resolve the search.
parameters_resolver	parameters to be given to the index in order to resolve the search.
free_parameters_on_resolver_release	true if `parameters_resolver` must be released during mknn_resolver_release.

Returns: an object configured to resolve a similarity search.

void mknn_index_release ( MknnIndex * index )

Releases the index.

Parameters

index the index to be released.

MknnIndex* mknn_index_restore	(	const char *	filename_read,
		MknnDataset *	search_dataset,
		bool	free_search_dataset_on_index_release,
		MknnDistance *	distance,
		bool	free_distance_on_index_release,
		MknnIndexParams *	more_parameters,
		bool	free_parameters_on_index_release
	)

Loads an index from a file.

The file must have been created by mknn_index_save.

Remarks: search_dataset and distance must be the same that were used to create and save the index.

Parameters

filename_read	File to read. If the file does not exists an error is raised.
search_dataset	The search dataset that was used to build the index.
free_search_dataset_on_index_release	binds the lifetime of `search_dataset` to the new index.
distance	The distance that was used to build the index.
free_distance_on_index_release	binds the lifetime of `distance` to the new index.
more_parameters	any parameter that may override the stored parameters or new parameters that could not be saved.
free_parameters_on_index_release	binds the lifetime of `more_parameters` to the new index.

Returns: a new index (it must be released with mknn_index_release).

void mknn_index_save	(	MknnIndex *	index,
		const char *	filename_write
	)

Saves the built index to a file.

It may create other files using filename_write as prefix. The created files may be in binary format for fast loading.

Remarks: The search_dataset and the distance are not saved.

Parameters

filename_write	File to create. If the file already exists it is overwritten.
index	the index to save

Functions

Detailed Description

Function Documentation