phrases-main.c

Ripper 分类算法

#include <stdio.h>
#include "ripper.h"
#include "protos.h"
#include "mdb.h"


/******************************************************************************/
char *Program="phrases";
char *Help_str[] = {
    "syntax: phrases [options] [stem]",
    "   add phrases to a set-valued fields of a dataset",
    "",
    "options are:",
    "  -s:       read from stdin",
    "  -n <str>: use attribute named <str>", 
    "  -D #:     max distance is #", 
    "  -N yn:    add 'near' facts (default y)",
    "  -A yn:    add 'after' facts (default y)",
    NULL
};

main(argc,argv)
int argc;
char *argv[];
{
    vec_t *data;
    char *stem;
    BOOL use_stdin,add_near,add_after;
    int i,j,k,el,dist,end;
    example_t *exi;
    symbol_t *wj,*wk,*wnew;
    char cbuf[BUFSIZ];
    int o;
    char *attr_name;
    int attr_index;
    vec_t *seti;
    int oldmax;

    /* defaults */
    set_trace_level(SUMM);
    use_stdin = FALSE;
    add_near = add_after = TRUE;
    dist = 2;
    attr_name = "";

    while ((o=getopt(argc,argv,"sm:D:N:A:h"))!=EOF) {
	switch (o) {
	  case 's':
	    use_stdin = TRUE;
	    break;
	  case 'D':
	    dist = atoi(optarg);
	    break;
	  case 'N':
	    add_near = (optarg[0]=='y' || optarg[0]=='Y');
	    break;
	  case 'A':
	    add_after = (optarg[0]=='y' || optarg[0]=='Y');
	    break;
	  case 'n':
	    attr_name = optarg;
	    break;
	  case 'h':
	  case '?':
	  default: 
	    give_help();
	    if (o=='h') exit(0);
	    else fatal("option not implemented");
	}
    }

    if (optind<argc) {
	stem = argv[optind++];
	ld_names(add_ext(stem,".names"));
	if (use_stdin) data = ld_data(NULL);
	else data = ld_data(add_ext(stem,".data"));
    } else {
	data = ld_data(NULL);
    }
    if (optind<argc) {
	warning("not all arguments were used: %s ...",argv[optind]);
    }

    if (data==NULL || vmax(data)==0) fatal("no data!");

    /* figure out which attribute is being used */
    attr_index = 0; /* default */
    for (i=0; i<n_fields(); i++) {
	if (strcmp(field_name(i),attr_name)==0) {
	    attr_index = i;
	    break;
	}
    }
    if (!set_field(attr_index)) {
	fatal("attribute %s is not set-valued",field_name(attr_index));
    }

    for (i=0; i<vmax(data); i++) {
	exi = vref(example_t,data,i);
	seti = vref(aval_t,exi->inst,attr_index)->u.set;

	/* add phrases to example */
	oldmax = vmax(seti);
	for (j=0; j<oldmax; j++) {
	    end = oldmax-1;
	    if (end >= j+dist) end = j+dist;
	    for (k=j+1; k<=end; k++) {
		wj = *vref(symbol_t *,seti,j);
		wk = *vref(symbol_t *,seti,k);
		if (add_near) {
		    for (el=k-j;el<=dist;el++) {
			if (strcmp(wj->name,wk->name) <= 0) {
			    sprintf(cbuf,"near%d(%s,%s)",
				   el,wj->name,wk->name);
			    wnew = intern(cbuf);
			    ext_vec(symbol_t *,seti,&wnew);
			} else {
			    sprintf(cbuf,"near%d(%s,%s)",
				   el,wj->name,wk->name);
			    wnew = intern(cbuf);
			    ext_vec(symbol_t *,seti,&wnew);
			}
		    }
		}
		if (add_after) {
		    sprintf(cbuf,"after%d(%s,%s)",
			    dist,wj->name,wk->name);
		    wnew = intern(cbuf);
		    ext_vec(symbol_t *,seti,&wnew);
		}
	    }/* for later word k*/
	}/*for earlier word j*/
    } /* for example i */

    /* print the extended dataset */
    print_data(data);
}