lshkit::Benchmark< KEY > Class Template Reference

Access a benchmark file. More...

#include <eval.h>

List of all members.

Public Member Functions
void	resize (unsigned Q, unsigned K=0)
void	init (unsigned Q, unsigned maxID)
void	load (std::istream &is)
void	save (std::ostream &os) const
void	load (const std::string &path)
void	save (const std::string &path) const
unsigned	getQ () const
unsigned	getQuery (unsigned n) const
	Get the ID of the nth query.
const Topk< KEY > &	getAnswer (unsigned n) const
	Get the nearest neighbors of the nth query.
Topk< KEY > &	getAnswer (unsigned n)
	Get the KNNs for modification.

---

Detailed Description

template<typename KEY = unsigned>
class lshkit::Benchmark< KEY >

Access a benchmark file.

We assume the feature vectors in the benchmark database are numbered from 0 to N. We sample Q queries as test examples and run K-NN search against the database with linear scan. The results are saved in a benchmark file for evaluation purpose. A benchmark file is made up of Q lines, each line represents a test query and is of the following format:

<query id>=""> <K> <1st NN's ID> <distance> <2nd NN's ID> <distance> ... <Kth NN's ID> <distance>

For all queries in the benchmark file, the K value should be the same.

Because the query points are also sampled from the database, they should be excluded from scanning when running this particular query.

---

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

• include/lshkit/eval.h

---