AbstractStringFeature Class Reference

#include <AbstractString.h>

Inheritance diagram for AbstractStringFeature:

List of all members.

Public Types
enum	ComparatorType { None, Levenshtein }
Public Member Functions
virtual double	getPosition () const
virtual double	getDistance (Feature *f) const
virtual void	moveTowards (Feature *f, double factor)
virtual string	serialize () const
virtual void	unserialize (string value)
virtual featureparams	write () const
virtual void	read (featureparams *param)
virtual Feature *	clone () const
virtual void	aggregate (aggregatelist samples)
virtual FeatureType	getType () const
virtual const string	getName () const
const string &	getVal () const
const unsigned long	getCodeVal () const
const stringcode *	getCode ()
virtual string	toString () const
Protected Member Functions
	AbstractStringFeature (ComparatorType comp, stringcode *code, long *maxlen)
	AbstractStringFeature (ComparatorType comp, stringcode *code, long *maxlen, const string &name)
unsigned long	getCodeForName (const string &name)
Protected Attributes
long *	maxlength
stringcode *	code
char *	position
unsigned long	codeval
Related Functions
(Note that these are not member functions.)
static long	levenshtein (char **x, const char *t, double *factor)

---

Detailed Description

Baseclass for features providing string values

A feature of type nominal which has character strings as values. For comparing values of this feature, either the normal string comparion with the number of different characters (Manhattan distance) or the Levenshtein distance can be used.

---

Member Enumeration Documentation

enum AbstractStringFeature::ComparatorType

The distance metric to use

---

Constructor & Destructor Documentation

AbstractStringFeature::AbstractStringFeature (  ComparatorType  comp, stringcode *  code, long *  maxlen )  [protected]

Feature constructor This constructor initializes the feature as prototypes value. For creating a specific sample of a feature, the sample constructor should be used. Parameters: comp  The distance metric that should be used. code  A reference to a static variable (it must be same for all created objects of _one_ feature) for persistantly storing the code table. maxlen  A reference to a static variable (it must be same for all created objects of _one_ feature) for persistantly storing the maximum string length.

AbstractStringFeature::AbstractStringFeature (  ComparatorType  comp, stringcode *  code, long *  maxlen, const string &  name )  [protected]

Sample constructor This constructor initializes the feature with a specific value and should thus be used for samples. Parameters: comp  The distance metric that should be used. code  A reference to a static variable (it must be same for all created objects of _one_ feature) for persistantly storing the code table. maxlen  A reference to a static variable (it must be same for all created objects of _one_ feature) for persistantly storing the maximum string length. name  The string representation of the sample value.

---

Member Function Documentation

void AbstractStringFeature::aggregate (  aggregatelist  samples  )  [virtual]

Aggregate a sample values from other sources Parameters: samples  A list of <timestamp, sample> tuples. Implements Feature.

Feature * AbstractStringFeature::clone (   )  const [virtual]

Clone a feature A feature must be able to clone itself. This is typically implemented with a copy constructor, but should return a new object. Implements Feature. Reimplemented in AbstractStringListFeature.

unsigned long AbstractStringFeature::getCodeForName (  const string &  name  )  [protected]

Get a code value for a given string A helper function to return the respective code for a given string. If the string is encountered the first time a new code is allocated. Parameters: name  The string to code Returns: A code value

double AbstractStringFeature::getDistance (  Feature *  f  )  const [virtual]

Todo: implement a better 'get string for code' Implements Feature. Reimplemented in AbstractStringListFeature.

virtual const string AbstractStringFeature::getName (   )  const [inline, virtual]

Query a features name Returns: Name of the Feature in the format "Featureprovider.Feature" Implements Feature. Reimplemented in AbstractStringListFeature, ActiveWindowFeature, BluetoothPeersFeature, GSMCellFeature, SystemCommandStringListFeature, WlanActiveEssidFeature, WlanActiveMacAddressFeature, and WlanPeersFeature.

double AbstractStringFeature::getPosition (   )  const [virtual]

Query a features position Returns: The distance to the origin. Every implementation must take care that only values in the interval [0; 1] are returned. Remarks: This function is only for the purpose of internal search operations and visualisation and should not be used in any other context. Implements Feature. Reimplemented in AbstractStringListFeature.

virtual FeatureType AbstractStringFeature::getType (   )  const [inline, virtual]

Query a features type Returns: The type of the Feature. Implements Feature.

const string& AbstractStringFeature::getVal (   )  const [inline]

Query the string values Returns: Values list

void AbstractStringFeature::moveTowards (  Feature *  f, double  factor )  [virtual]

Todo: implement a better 'get string for code' Implements Feature. Reimplemented in AbstractStringListFeature.

void AbstractStringFeature::read (  featureparams *  param  )  [virtual]

Load feature from persistant data Initializes the persistant feature data from the given representation. Parameters: param  Persistant feature data. See also: write Implements PersistantFeature.

string AbstractStringFeature::serialize (   )  const [virtual]

Serialize a samples data to a string Returns: String representation of the samples data. Implements Feature. Reimplemented in AbstractStringListFeature.

string AbstractStringFeature::toString (   )  const [virtual]

Get feature as string Note: This is only for testing. Returns: Feature as string. Implements Feature. Reimplemented in AbstractStringListFeature, ActiveWindowFeature, BluetoothPeersFeature, GSMCellFeature, SystemCommandStringListFeature, WlanActiveEssidFeature, WlanActiveMacAddressFeature, and WlanPeersFeature.

void AbstractStringFeature::unserialize (  string  value  )  [virtual]

Unserialize a samples data from a string Parameters: value  String representation of the samples data. Implements Feature. Reimplemented in AbstractStringListFeature.

featureparams AbstractStringFeature::write (   )  const [virtual]

Externalize feature Returns: Persistant feature data. See also: read Implements PersistantFeature.

---

Friends And Related Function Documentation

static long levenshtein (  char **  x, const char *  t, double *  factor )  [related]

Modified Levenshtein algorithm This function implements a slightly modified version of the Levenshtein algorithm to not only calculate the distance of two strings but also modify the string x towards the string t with the propability factor. In other words, every transformation operation found by the algorithm is performed on the string x with the propability factor. Parameters: x  Input string. t  String to compare the input string to. factor  Propability with witch transformations are performed. Has to be a value out of the intervall [0; 1]. Returns: The levenshtein distance between x and t. Todo: alloc once

---

Member Data Documentation

stringcode* AbstractStringFeature::code [protected]

A reference to the static, persistant code table of strings seen so far.

unsigned long AbstractStringFeature::codeval [protected]

The coded value of the feature See also: name code

long* AbstractStringFeature::maxlength [protected]

A reference to the static, persistant maximum length of strings seen so far.

char* AbstractStringFeature::position [protected]

Only for internal usage in moveTowards and getDistance.

---

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

• amw/src/cpp/cross/contextprediction/framework/Features/AbstractString.h
• amw/src/cpp/cross/contextprediction/framework/Features/AbstractString.cpp

---