DetLemma Class Reference
[LeJa]

putting the pieces together for determining the lemma More...

#include <DetLemma.h>

List of all members.

Public Member Functions
	DetLemma ()
	default constructor
	~DetLemma ()
	destructor
bool	fillFlex (char *)
	initializes the levels from XML file
bool	detLemma (string &, JXDictionary *, list< result_lemma > &, FILE *)
	this function creates HypoLemma objects for possible segmentation of the hiragana ending
bool	confirmLemma (string &, string &, JXDictionary *, result_lemma &)
	searches a hypothetical lemma in the dictionary
bool	findInLevel (string &, UINT)
	searches a hiragana string in a specific level
bool	insertFlextoLevel (FlexLevel *, int)
	inserts a single Flexlevel into vector m_flexlevel
void	HypoPrint ()
	prints all hypothesis to stdout
bool	insertPossibleHypos (JVerbalFlexion *jvf, UMString hirastr, int j)
	this function adds on to the possible reduction hypothesis
Private Types
typedef vector< FlexLevel * >::iterator	IT
typedef vector< HypoLemma >::iterator	HIT
Private Member Functions
int	getHighestLevel (char *)
	finds the highest level in grammar
Private Attributes
vector< FlexLevel * >	m_flexlevel
vector< HypoLemma >	m_hypothesen
int	m_number_of_levels
int	m_number_of_entries
int	m_flexmaxlength
int	m_level_counter [MAXFLEX]
list< result_lemma >	m_result_list
string	m_cur_reqform
string	m_cur_reqflexart
JVerbalFlexion *	hypo

---

Detailed Description

putting the pieces together for determining the lemma

Author:

Iris Vogel iris@urz.uni-heidelberg.de

Date:

Aug 2003

Version:

0.8/15

See also:

DetLemma.h

---

Member Typedef Documentation

typedef vector<HypoLemma>::iterator DetLemma::HIT [private]

iterator for hyothesis vector

typedef vector<FlexLevel*>::iterator DetLemma::IT [private]

iterator for the flexation levels

---

Constructor & Destructor Documentation

DetLemma::DetLemma (   )

default constructor

DetLemma::~DetLemma (   )

destructor iterates through the map of flexlevels and deletes each one

---

Member Function Documentation

bool DetLemma::confirmLemma (  string &  lemma, string &  flexart, JXDictionary *  jxdic, result_lemma &  rl )

searches a hypothetical lemma in the dictionary by not only checking the lemma-string but also the kind of flexation, results are good. however, there can be more than one result for the same lookup, especially when the initial string is an all-hiragana string Parameters: lemma  lemma in question flexart  kind of flexation jxdic  pointer JXDictionary class for lookup rl  variable for returning the result, if lookup succeeded Returns: true if successful false if unsuccessful

bool DetLemma::detLemma (  string &  sstring, JXDictionary *  jxdic, list< result_lemma > &  result_list, FILE *  logfile )

this function creates HypoLemma objects for possible segmentation of the hiragana ending This is the heart of lemmatization. First there is the hiragana string, every possible way to split it results in a new hypothesis about stem and ending. For every hypothesis the ending is searched in the grammar, starting at the highest level. Once an ending is found, it gets cut of and the rest of the hiragana string is treated again as above. Once all the levels of flexation and hypothesis are checked, the dictionary form ending of the verb is selcted from the grammar, agglutinated and checked against the dictionary. Also the kind of verb associated with the grammatical rules is checked to make sure that similar forms like passiv and potential are separated if possible. Parameters: sstring  full verbal string to be lemmatized jxdic  dictionary class pointer for lookup result_list  list of (references) to result of lemmatization and lookup Returns: true if lookup of possible lemmas succeeded false if lookup possible lemmas failed Todo: it would be good (save a lot of .GetStringUC()-calls) to overload this function with a UNICODE string oder wchar_t* instead of the multibyte string

bool DetLemma::fillFlex (  char *  fname  )

initializes the levels from XML file this function is responsible for the grammar input the grammatical verbforms are hardcoded, which is no big deal if you stick to Japanese, because the numbers and kinds are very much fixed it prints a list of the loaded Flexlevels to stdout Parameters: fname  name of the grammar file Returns: true, if succesful

bool DetLemma::findInLevel (  string &  end, UINT  level )

searches a hiragana string in a specific level Parameters: end  hiragana ending in question level  level that is being searched Returns: true if successful false if unsuccessful

int DetLemma::getHighestLevel (  char *   )  [private]

finds the highest level in grammar internal funktion to determine how many levels

void DetLemma::HypoPrint (   )

prints all hypothesis to stdout

bool DetLemma::insertFlextoLevel (  FlexLevel *  fl, int  pos )

inserts a single Flexlevel into vector m_flexlevel Parameters: fl  : name of the grammar file pos  : name for Returns: true, if succesful

bool DetLemma::insertPossibleHypos (  JVerbalFlexion *  jvf, UMString  hirastr, int  j )

this function adds on to the possible reduction hypothesis this fuction is called when changes to the hiragana string occured, especially when an ending has been recognized and deleted from the string. the new, shorter string is being split into parts again, and for each new possibilty an hypothesis is added Parameters: jvf  flexation class of hypothesis in question jhirastr  Hiragana string that is going to split at point specified by j j  splitting point for Hiragana string Returns: true if successful false if unsuccessful

---

Member Data Documentation

JVerbalFlexion* DetLemma::hypo [private]

pointer for processing current hypothesis

string DetLemma::m_cur_reqflexart [private]

variable for processing current hypothesis

string DetLemma::m_cur_reqform [private]

variable for processing current hypothesis

vector<FlexLevel*> DetLemma::m_flexlevel [private]

the flexation levels of the grammar file are stored here

int DetLemma::m_flexmaxlength [private]

variable, that defines the maximum length of an grammatical unit of hiragana

vector<HypoLemma> DetLemma::m_hypothesen [private]

hypthesis for lemma resolution are stored here

int DetLemma::m_level_counter[MAXFLEX] [private]

array storing the numbers corresponding to the flexlevels

int DetLemma::m_number_of_entries [private]

variable, that defines the number of entries

int DetLemma::m_number_of_levels [private]

variable, that stores the number of levels found in the grammar file

list<result_lemma> DetLemma::m_result_list [private]

since sometimes there is more than one resolution, results are stored in a list

---

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

• DetLemma.h
• DetLemma.cpp

---