E
- the type of MTreeEntry used in the M-TreeN
- the type of AbstractMTreeNode used in the M-Tree@Priority(value=200) @Reference(authors="P. Ciaccia, M. Patella, P. Zezula", title="M-tree: An Efficient Access Method for Similarity Search in Metric Spaces", booktitle="Proc. Int. Conf. Very Large Data Bases (VLDB\'97)", url="http://www.vldb.org/conf/1997/P426.PDF", bibkey="DBLP:conf/vldb/CiacciaPZ97") @Alias(value="de.lmu.ifi.dbs.elki.index.tree.metrical.mtreevariants.split.MLBDistSplit") public class MLBDistSplit<E extends MTreeEntry,N extends AbstractMTreeNode<?,N,E>> extends AbstractMTreeSplit<E,N>
The benefit of this strategy is that it works with precomputed distances from the parent, while most other strategies would require O(n²) distance computations. So if construction time is critical, this is a good choice.
Reference:
P. Ciaccia, M. Patella, P. Zezula
M-tree: An Efficient Access Method for Similarity Search in Metric
Spaces
Proc. Int. Conf. Very Large Data Bases (VLDB'97)
Modifier and Type | Class and Description |
---|---|
static class |
MLBDistSplit.Parameterizer<E extends MTreeEntry,N extends AbstractMTreeNode<?,N,E>>
Parameterization class.
|
distributor
Constructor and Description |
---|
MLBDistSplit(DistributionStrategy distributor)
Constructor.
|
Modifier and Type | Method and Description |
---|---|
Assignments<E> |
split(AbstractMTree<?,N,E,?> tree,
N node)
Selects the second object of the specified node to be promoted and stored
into the parent node and partitions the entries according to the M_LB_DIST
strategy.
|
computeDistanceMatrix
public MLBDistSplit(DistributionStrategy distributor)
distributor
- Distribution strategypublic Assignments<E> split(AbstractMTree<?,N,E,?> tree, N node)
This strategy considers all possible pairs of objects and chooses the pair of objects for which the distance is maximum.
tree
- Tree to usenode
- the node to be splitCopyright © 2019 ELKI Development Team. License information.