prob.c

中心词驱动的短语结构句法分析器。该模型考虑了跟随介词短语的名词短语的中心词的作用。 有MIT大学Colling开发

  5           punc
  6           cc

  7           p
  8           ch
  9           dist
  10..12      subcat
  13          th
  14..15      wh

*/

void calc_newdist(int *newd,int d)
{
  if(d>=100)
    {
      *newd = 100;
      d-=100;
    }
  else
    *newd=0;

  if(d>=10)
    {
      if(DISTAFLAG)
	*newd += 10;
      d-=10;
    }

  if(d>=1)
    {
      if(DISTVFLAG)
	*newd += 1;
    }
}

void make_dep1_string(unsigned char *string,int wm,int tm,int cm,int wh,int th,int p,int ch,int dist,int subcat,int cc,int punc)
{
  int newdist;

  calc_newdist(&newdist,dist);

  string[3] = (char) tm;
  string[4] = (char) cm;
  string[5] = (char) punc;
  string[6] = (char) cc;

  string[7] = (char) p;
  string[8] = (char) ch;
  string[9] = (char) newdist;
  byte3_to_char(&string[10],subcat);
  string[13] = (char) th;
  byte2_to_char(&string[14],wh);

}

/*make string for P(wm | cm,t, context) 

  position    element
  3..4        wm

  5           tm
  6           cm
  7           punc
  8           cc
  9           p
  10           ch
  11          dist
  12..14      subcat
  15          th
  16..17      wh
*/

void make_dep2_string(unsigned char *string,int wm,int tm,int cm,int wh,int th,int p,int ch,int dist,int subcat,int cc,int punc)
{
  int newdist;

  calc_newdist(&newdist,dist);

  byte2_to_char(&string[3],wm);

  string[5] = (char) tm;
  string[6] = (char) cm;
  string[7] = (char) punc;
  string[8] = (char) cc;

  string[9] = (char) p;
  string[10] = (char) ch;
  string[11] = (char) newdist;
  byte3_to_char(&string[12],subcat);
  string[15] = (char) th;
  byte2_to_char(&string[16],wh);

}

/*make string for P(wm | cm,t, context) 

  position    element
  3..4        wm

  5           tm
*/

void add_tagword_counts(int wm,int tm,hash_table *hash)
{
  unsigned char buffer[1000];
  int bos[] = {1,1};

  wm = fwords[wm];

  byte2_to_char(&buffer[3],wm);

  buffer[5] = (char) tm;

  add_counts_level(buffer,D2OLEN,bos,3,D2PROBTYPE,hash);
  add_counts_level(buffer,S2OLEN,bos,2,S2PROBTYPE,hash);
  add_counts_level(buffer,CP2OLEN,bos,3,CP2PROBTYPE,hash);
}


void byte2_to_char(unsigned char *string,int n)
{
  string[0] = (n & 255);
  string[1] = n/256;
}

void byte3_to_char(unsigned char *string,int n)
{
  string[0] = (n&255);
  string[1] = ((n/256)&255);
  string[2] = n/65536;
}


/*make string for P(ch | p,th,wh)

  position    element
  3           ch

  4           p
  5           th
  6..7        wh
*/

void make_unary_string(unsigned char *string,int ch,int wh,int th,int p)
{

  string[3] = (char) ch;

  string[4] = (char) p;
  string[5] = (char) th;
  byte2_to_char(&string[6],wh);
}

/*make string for P(ch,th | p==TOP)

  position    element
  3           ch
  4           th

  5           TOP

*/

void make_s1_string(unsigned char *string,int ch,int wh,int th,int p)
{

  string[3] = (char) ch;
  string[4] = (char) th;

  string[5] = (char) p;
}

/*make string for P(wh | ch,th, p==TOP)

  position    element
  3..4        wh

  5           th
  6           ch
  7           TOP

*/

void make_s2_string(unsigned char *string,int ch,int wh,int th,int p)
{

  byte2_to_char(&string[3],wh);

  string[5] = (char) th;
  string[6] = (char) ch;
  string[7] = (char) p;
}

/*make string for P(subcat | wh,ch,th, p,dir)

  position    element
  3..5        subcat

  6           ch
  7           p
  8           dir
  9           th
  10..11      wh

*/

void make_subcat_string(unsigned char *string,int subcat,int ch,int wh,int th,int p,int dir)
{
  byte3_to_char(&string[3],subcat);

  string[6] = (char) ch;
  string[7] = (char) p;
  string[8] = (char) dir;
  string[9] = (char) th;
  byte2_to_char(&string[10],wh);
}

/*make string for P(ch | wh,th)

  position    element
  3           ch

  4           0
  5           th
  6..7        wh

*/

void make_prior1_string(unsigned char *string,int ch,int wh,int th)
{

  string[3] = (char) ch;

  string[4] = 0;
  string[5] = (char) th;
  byte2_to_char(&string[6],wh);
}

/*make string for P(wh, th |  p==anything)

  position    element
  3..4        wh
  5           th

  6           0

*/

void make_prior2_string(unsigned char *string,int ch,int wh,int th)
{

  byte2_to_char(&string[3],wh);
  string[5] = (char) th;

  string[6] = 0;

}

/* pcc1_string

   position      element
   3             t

   4             type
   5             p
   6             ch
   7             cm
   8             th
   9             tm
   10            wh
   12            wm
*/

void make_pcc1_string(unsigned char *string,int t,int p,int ch,int th,int wh,int cm,int tm,int wm,int type)
{
/*  printf("PCC1 %d %d %d %d %d %d %d %d %d\n",
	 t,p,ch,th,wh,cm,tm,wm,type);*/


  string[3] = (char) t;

  string[4] = (char) type;

  string[5] = (char) p;
  string[6] = (char) ch;
  string[7] = (char) cm;

  string[8] = (char) th;
  string[9] = (char) tm;

  byte2_to_char(&string[10],wh);
  byte2_to_char(&string[12],wm);
}

/* pcc2_string

   position      element
   3..4          w

   5             t
   6             type
   7             p
   8             ch
   9             cm
   10            th
   11            tm
   12            wh
   14            wm
*/

void make_pcc2_string(unsigned char *string,int w,int t,int p,int ch,int th,int wh,int cm,int tm,int wm,int type)
{

/*  printf("PCC1 %d %d %d %d %d %d %d %d %d %d\n",
	 w,t,p,ch,th,wh,cm,tm,wm,type);*/

  byte2_to_char(&string[3],w);
  
  string[5] = (char) t;

  string[6] = (char) type;

  string[7] = (char) p;
  string[8] = (char) ch;
  string[9] = (char) cm;

  string[10] = (char) th;
  string[11] = (char) tm;

  byte2_to_char(&string[12],wh);
  byte2_to_char(&string[14],wm);
}


/*make string for P(gap | wh,ch,th, p)

  position    element
  3           gap

  4           ch
  5           p
  6           th
  7..8      wh

*/

void make_gap_string(unsigned char *string,int gap,int ch,int wh,int th,int p)
{
  string[3] = (char) gap;

  string[4] = (char) ch;
  string[5] = (char) p;
  string[6] = (char) th;
  byte2_to_char(&string[7],wh);
}