quickrank::data::Dataset Class Reference

This class implements a Dataset to be used for a L-t-R task. More...

#include <dataset.h>

Public Member Functions
	Dataset (size_t n_instances, size_t n_features)
	Allocates an empty Dataset of given size in horizontal format. More...
virtual	~Dataset ()
	Dataset (const Dataset &other)=delete
	Avoid inefficient copy constructor. More...
Dataset &	operator= (const Dataset &)=delete
	Avoid inefficient copy assignment. More...
quickrank::Feature *	at (size_t document_id, size_t feature_id)
	Returns a pointer to a specific data item. More...
Label	getLabel (size_t document_id)
	Returns the value of the i-th relevance label. More...
size_t	offset (size_t i) const
	Returns the offset in the internal data structure of the i-th query results list. More...
std::unique_ptr< QueryResults >	getQueryResults (size_t i) const
	Returns the i-th QueryResults in the dataset. More...
void	addInstance (QueryID q_id, Label i_label, std::vector< Feature > i_features)
	Add a new training instance, i.e., a labeled document, to the dataset. More...
size_t	num_features () const
	Returns the number of features used to represent a document. More...
size_t	num_queries () const
	Returns the number of queries in the dataset. More...
size_t	num_instances () const
	Returns the number of documents in the dataset. More...

Private Member Functions
virtual std::ostream &	put (std::ostream &os) const
	Prints the data reading time stats. More...

Private Attributes
size_t	num_features_
size_t	num_queries_
size_t	num_instances_
quickrank::Feature *	data_ = NULL
quickrank::Label *	labels_ = NULL
std::vector< size_t >	offsets_
size_t	last_instance_id_
size_t	max_instances_

Friends
std::ostream &	operator<< (std::ostream &os, const Dataset &me)
	The output stream operator. More...

Detailed Description

This class implements a Dataset to be used for a L-t-R task.

The internal representation is quite simple: a row vector of size num_instances() x num_features(). (A training instance is indeed a document.) We allow to directly access the internal representation through the function at() to support fast access and custom high performance implementations. Internal representation is horizontal (instances x features).

Constructor & Destructor Documentation

quickrank::data::Dataset::Dataset	(	size_t	n_instances,
		size_t	n_features
	)

Allocates an empty Dataset of given size in horizontal format.

Parameters

n_instances	The number of training instances (lines) in the dataset.
n_features	The number of features.

quickrank::data::Dataset::~Dataset ( )

virtual

quickrank::data::Dataset::Dataset ( const Dataset &  other )

delete

Avoid inefficient copy constructor.

Member Function Documentation

void quickrank::data::Dataset::addInstance	(	QueryID	q_id,
		Label	i_label,
		std::vector< Feature >	i_features
	)

Add a new training instance, i.e., a labeled document, to the dataset.

Warning

Currently the addition works only when data is in HORIZ format.

Parameters

q_id	The query ID.
i_label	The relevance label of the result.
i_features	The feature vector of the document.

quickrank::Feature* quickrank::data::Dataset::at ( size_t  document_id, size_t  feature_id  )

inline

Returns a pointer to a specific data item.

Parameters

document_id	The document of interest.
feature_id	The feature of interest.

Returns

A reference to the requested feature value of the given document id.

Label quickrank::data::Dataset::getLabel ( size_t  document_id )

inline

Returns the value of the i-th relevance label.

std::unique_ptr< QueryResults > quickrank::data::Dataset::getQueryResults

(

size_t

i

)

const

Returns the i-th QueryResults in the dataset.

Parameters

i	The i-th query results list of interest.

Returns

The requested QueryResults.

size_t quickrank::data::Dataset::num_features ( ) const

inline

Returns the number of features used to represent a document.

size_t quickrank::data::Dataset::num_instances ( ) const

inline

Returns the number of documents in the dataset.

size_t quickrank::data::Dataset::num_queries ( ) const

inline

Returns the number of queries in the dataset.

size_t quickrank::data::Dataset::offset ( size_t  i ) const

inline

Returns the offset in the internal data structure of the i-th query results list.

Parameters

i	The i-th query results list of interest.

Returns

The offset of the first document in the i-th query results list. This can be used to later invoke the at() function.

Dataset& quickrank::data::Dataset::operator= ( const Dataset &  )

delete

Avoid inefficient copy assignment.

std::ostream & quickrank::data::Dataset::put ( std::ostream &  os ) const

privatevirtual

Prints the data reading time stats.

Friends And Related Function Documentation

std::ostream& operator<< ( std::ostream &  os, const Dataset &  me  )

friend

The output stream operator.

Prints the data reading time stats

Member Data Documentation

quickrank::Feature* quickrank::data::Dataset::data_ = NULL

private

quickrank::Label* quickrank::data::Dataset::labels_ = NULL

private

size_t quickrank::data::Dataset::last_instance_id_

private

size_t quickrank::data::Dataset::max_instances_

private

size_t quickrank::data::Dataset::num_features_

private

size_t quickrank::data::Dataset::num_instances_

private

size_t quickrank::data::Dataset::num_queries_

private

std::vector<size_t> quickrank::data::Dataset::offsets_

private

---

The documentation for this class was generated from the following files:

• include/data/dataset.h
• src/data/dataset.cc