---

ClusterDB Class Reference

Abstract interface for clustering databases. More...

#include <ClusterDB.hpp>

Inheritance diagram for ClusterDB:

List of all members.

Public Methods
	ClusterDB (const Index *ind, double threshold=0.25, enum ClusterParam::simTypes simType=ClusterParam::COS, enum ClusterParam::clusterTypes clusterType=ClusterParam::CENTROID, enum ClusterParam::docModes docMode=ClusterParam::DMAX)
	initialize the cluster method
virtual	~ClusterDB ()
	clean up
virtual int	countClusters () const=0
	Return number of clusters.
virtual int	maxID () const=0
	Return highest cluster ID.
virtual vector< Cluster * >	getDocCluster (DOCID_T docId) const=0
	Get the Cluster for the given docId.
virtual Cluster *	getCluster (int clusterId) const=0
	Get the Cluster for the given clusterId.
virtual vector< int >	getDocClusterId (DOCID_T docId) const=0
	Get the Cluster id for the given docId.
virtual int	addToCluster (DOCID_T docId, int clusterId, double score)=0
	Add a document id to a cluster, given the cluster id.
virtual int	addToCluster (DOCID_T docId, Cluster *cluster, double score)=0
	Add a document id to a cluster, given the cluster.
virtual int	removeFromCluster (DOCID_T docId, int clusterID)=0
	Remove a document id from a cluster, given the cluster id.
virtual int	deleteCluster (int clusterID)=0
	Delete a cluster entirely.
virtual int	deleteCluster (Cluster *target)=0
	Delete a cluster entirely.
int	addCluster (Cluster *oldCluster)
	Add a cluster to the DB (renumbers the cluster id). Returns new id.
virtual vector< int >	splitCluster (int cid, int num=2)=0
	split cluster using Cluster::split
virtual int	mergeClusters (int cid1, int cid2)=0
	merge two clusters.
virtual void	printClusters () const
	pretty print all clusters to the standard output.
virtual int	cluster (DOCID_T docId)
	Assign a document to a cluster.
virtual int	cluster (DOCID_T docId, double &finalScore)
	Assign a document to a cluster, returning score.
virtual string	getKeyWords (int cid, int numTerms=10) const
	Get the top N keywords for a cluster.
Protected Methods
virtual Cluster *	newCluster ()=0
	subclass specific cluster intitialization.
Cluster *	allocateCluster (int clusterID) const
	Uses ClusterFactory to create Cluster objects.
Protected Attributes
const Index *	index
	Database containing the collection to operate on.
int	numDocs
	Number of documents in the database, reduces calls to db->docCount().
int	numTerms
	Number of terms in the database.
double	threshold
	threshold for YES/NO decisions
vector< Cluster * >	clusters
	the cluster database.
const SimilarityMethod *	sim
	Similarity method to use.
ClusterFactory *	factory
	Cluster factory.
ThresholdFcn *	thresh
	Threshold function for adaptive thresholding.

---

Detailed Description

Abstract interface for clustering databases.

---

Constructor & Destructor Documentation

ClusterDB::ClusterDB (  const Index *    ind, double    threshold = 0.25, enum ClusterParam::simTypes    simType = ClusterParam::COS, enum ClusterParam::clusterTypes    clusterType = ClusterParam::CENTROID, enum ClusterParam::docModes    docMode = ClusterParam::DMAX )

initialize the cluster method

ClusterDB::~ClusterDB (    )  [virtual]

clean up

---

Member Function Documentation

int ClusterDB::addCluster (  Cluster *    oldCluster )

Add a cluster to the DB (renumbers the cluster id). Returns new id.

virtual int ClusterDB::addToCluster (  DOCID_T    docId, Cluster *    cluster, double    score )  [pure virtual]

Add a document id to a cluster, given the cluster. Implemented in FlatFileClusterDB, and KeyfileClusterDB.

virtual int ClusterDB::addToCluster (  DOCID_T    docId, int    clusterId, double    score )  [pure virtual]

Add a document id to a cluster, given the cluster id. Implemented in FlatFileClusterDB, and KeyfileClusterDB.

Cluster * ClusterDB::allocateCluster (  int    clusterID )  const [protected]

Uses ClusterFactory to create Cluster objects.

int ClusterDB::cluster (  DOCID_T    docId, double &    finalScore )  [virtual]

Assign a document to a cluster, returning score.

int ClusterDB::cluster (  DOCID_T    docId )  [virtual]

Assign a document to a cluster.

virtual int ClusterDB::countClusters (    )  [pure virtual]

Return number of clusters. Implemented in FlatFileClusterDB, and KeyfileClusterDB.

virtual int ClusterDB::deleteCluster (  Cluster *    target )  [pure virtual]

Delete a cluster entirely. Implemented in FlatFileClusterDB, and KeyfileClusterDB.

virtual int ClusterDB::deleteCluster (  int    clusterID )  [pure virtual]

Delete a cluster entirely. Implemented in FlatFileClusterDB, and KeyfileClusterDB.

virtual Cluster* ClusterDB::getCluster (  int    clusterId )  const [pure virtual]

Get the Cluster for the given clusterId. Implemented in FlatFileClusterDB, and KeyfileClusterDB.

virtual vector<Cluster*> ClusterDB::getDocCluster (  DOCID_T    docId )  const [pure virtual]

Get the Cluster for the given docId. Implemented in FlatFileClusterDB, and KeyfileClusterDB.

virtual vector<int> ClusterDB::getDocClusterId (  DOCID_T    docId )  const [pure virtual]

Get the Cluster id for the given docId. Implemented in FlatFileClusterDB, and KeyfileClusterDB.

string ClusterDB::getKeyWords (  int    cid, int    numTerms = 10 )  const [virtual]

Get the top N keywords for a cluster.

virtual int ClusterDB::maxID (    )  [pure virtual]

Return highest cluster ID. Implemented in FlatFileClusterDB, and KeyfileClusterDB.

virtual int ClusterDB::mergeClusters (  int    cid1, int    cid2 )  [pure virtual]

merge two clusters. Implemented in FlatFileClusterDB, and KeyfileClusterDB.

virtual Cluster* ClusterDB::newCluster (    )  [protected, pure virtual]

subclass specific cluster intitialization. Implemented in FlatFileClusterDB, and KeyfileClusterDB.

void ClusterDB::printClusters (    )  [virtual]

pretty print all clusters to the standard output.

virtual int ClusterDB::removeFromCluster (  DOCID_T    docId, int    clusterID )  [pure virtual]

Remove a document id from a cluster, given the cluster id. Implemented in FlatFileClusterDB, and KeyfileClusterDB.

virtual vector<int> ClusterDB::splitCluster (  int    cid, int    num = 2 )  [pure virtual]

split cluster using Cluster::split Implemented in FlatFileClusterDB, and KeyfileClusterDB.

---

Member Data Documentation

vector<Cluster *> ClusterDB::clusters [protected]

the cluster database.

ClusterFactory* ClusterDB::factory [protected]

Cluster factory.

const Index* ClusterDB::index [protected]

Database containing the collection to operate on.

int ClusterDB::numDocs [protected]

Number of documents in the database, reduces calls to db->docCount().

int ClusterDB::numTerms [protected]

Number of terms in the database.

const SimilarityMethod* ClusterDB::sim [protected]

Similarity method to use.

ThresholdFcn* ClusterDB::thresh [protected]

Threshold function for adaptive thresholding.

double ClusterDB::threshold [protected]

threshold for YES/NO decisions

---

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

• ClusterDB.hpp
• ClusterDB.cpp

---