wolffd@0: /***********************************************************************/
wolffd@0: /*                                                                     */
wolffd@0: /*   svm_classify.c                                                    */
wolffd@0: /*                                                                     */
wolffd@0: /*   Classification module of Support Vector Machine.                  */
wolffd@0: /*                                                                     */
wolffd@0: /*   Author: Thorsten Joachims                                         */
wolffd@0: /*   Date: 02.07.02                                                    */
wolffd@0: /*                                                                     */
wolffd@0: /*   Copyright (c) 2002  Thorsten Joachims - All rights reserved       */
wolffd@0: /*                                                                     */
wolffd@0: /*   This software is available for non-commercial use only. It must   */
wolffd@0: /*   not be modified and distributed without prior permission of the   */
wolffd@0: /*   author. The author is not responsible for implications from the   */
wolffd@0: /*   use of this software.                                             */
wolffd@0: /*                                                                     */
wolffd@0: /************************************************************************/
wolffd@0: 
wolffd@0: # include "svm_common.h"
wolffd@0: 
wolffd@0: char docfile[200];
wolffd@0: char modelfile[200];
wolffd@0: char predictionsfile[200];
wolffd@0: 
wolffd@0: void read_input_parameters(int, char **, char *, char *, char *, long *, 
wolffd@0: 			   long *);
wolffd@0: void print_help(void);
wolffd@0: 
wolffd@0: 
wolffd@0: int main (int argc, char* argv[])
wolffd@0: {
wolffd@0:   DOC *doc;   /* test example */
wolffd@0:   WORD *words;
wolffd@0:   long max_docs,max_words_doc,lld;
wolffd@0:   long totdoc=0,queryid,slackid;
wolffd@0:   long correct=0,incorrect=0,no_accuracy=0;
wolffd@0:   long res_a=0,res_b=0,res_c=0,res_d=0,wnum,pred_format;
wolffd@0:   long j;
wolffd@0:   double t1,runtime=0;
wolffd@0:   double dist,doc_label,costfactor;
wolffd@0:   char *line,*comment; 
wolffd@0:   FILE *predfl,*docfl;
wolffd@0:   MODEL *model; 
wolffd@0: 
wolffd@0:   read_input_parameters(argc,argv,docfile,modelfile,predictionsfile,
wolffd@0: 			&verbosity,&pred_format);
wolffd@0: 
wolffd@0:   nol_ll(docfile,&max_docs,&max_words_doc,&lld); /* scan size of input file */
wolffd@0:   max_words_doc+=2;
wolffd@0:   lld+=2;
wolffd@0: 
wolffd@0:   line = (char *)my_malloc(sizeof(char)*lld);
wolffd@0:   words = (WORD *)my_malloc(sizeof(WORD)*(max_words_doc+10));
wolffd@0: 
wolffd@0:   model=read_model(modelfile);
wolffd@0: 
wolffd@0:   if(model->kernel_parm.kernel_type == 0) { /* linear kernel */
wolffd@0:     /* compute weight vector */
wolffd@0:     add_weight_vector_to_linear_model(model);
wolffd@0:   }
wolffd@0:   
wolffd@0:   if(verbosity>=2) {
wolffd@0:     printf("Classifying test examples.."); fflush(stdout);
wolffd@0:   }
wolffd@0: 
wolffd@0:   if ((docfl = fopen (docfile, "r")) == NULL)
wolffd@0:   { perror (docfile); exit (1); }
wolffd@0:   if ((predfl = fopen (predictionsfile, "w")) == NULL)
wolffd@0:   { perror (predictionsfile); exit (1); }
wolffd@0: 
wolffd@0:   while((!feof(docfl)) && fgets(line,(int)lld,docfl)) {
wolffd@0:     if(line[0] == '#') continue;  /* line contains comments */
wolffd@0:     parse_document(line,words,&doc_label,&queryid,&slackid,&costfactor,&wnum,
wolffd@0: 		   max_words_doc,&comment);
wolffd@0:     totdoc++;
wolffd@0:     if(model->kernel_parm.kernel_type == 0) {   /* linear kernel */
wolffd@0:       for(j=0;(words[j]).wnum != 0;j++) {  /* Check if feature numbers   */
wolffd@0: 	if((words[j]).wnum>model->totwords) /* are not larger than in     */
wolffd@0: 	  (words[j]).wnum=0;               /* model. Remove feature if   */
wolffd@0:       }                                        /* necessary.                 */
wolffd@0:       doc = create_example(-1,0,0,0.0,create_svector(words,comment,1.0));
wolffd@0:       t1=get_runtime();
wolffd@0:       dist=classify_example_linear(model,doc);
wolffd@0:       runtime+=(get_runtime()-t1);
wolffd@0:       free_example(doc,1);
wolffd@0:     }
wolffd@0:     else {                             /* non-linear kernel */
wolffd@0:       doc = create_example(-1,0,0,0.0,create_svector(words,comment,1.0));
wolffd@0:       t1=get_runtime();
wolffd@0:       dist=classify_example(model,doc);
wolffd@0:       runtime+=(get_runtime()-t1);
wolffd@0:       free_example(doc,1);
wolffd@0:     }
wolffd@0:     if(dist>0) {
wolffd@0:       if(pred_format==0) { /* old weired output format */
wolffd@0: 	fprintf(predfl,"%.8g:+1 %.8g:-1\n",dist,-dist);
wolffd@0:       }
wolffd@0:       if(doc_label>0) correct++; else incorrect++;
wolffd@0:       if(doc_label>0) res_a++; else res_b++;
wolffd@0:     }
wolffd@0:     else {
wolffd@0:       if(pred_format==0) { /* old weired output format */
wolffd@0: 	fprintf(predfl,"%.8g:-1 %.8g:+1\n",-dist,dist);
wolffd@0:       }
wolffd@0:       if(doc_label<0) correct++; else incorrect++;
wolffd@0:       if(doc_label>0) res_c++; else res_d++;
wolffd@0:     }
wolffd@0:     if(pred_format==1) { /* output the value of decision function */
wolffd@0:       fprintf(predfl,"%.8g\n",dist);
wolffd@0:     }
wolffd@0:     if((int)(0.01+(doc_label*doc_label)) != 1) 
wolffd@0:       { no_accuracy=1; } /* test data is not binary labeled */
wolffd@0:     if(verbosity>=2) {
wolffd@0:       if(totdoc % 100 == 0) {
wolffd@0: 	printf("%ld..",totdoc); fflush(stdout);
wolffd@0:       }
wolffd@0:     }
wolffd@0:   }  
wolffd@0:   free(line);
wolffd@0:   free(words);
wolffd@0:   free_model(model,1);
wolffd@0: 
wolffd@0:   if(verbosity>=2) {
wolffd@0:     printf("done\n");
wolffd@0: 
wolffd@0: /*   Note by Gary Boone                     Date: 29 April 2000        */
wolffd@0: /*      o Timing is inaccurate. The timer has 0.01 second resolution.  */
wolffd@0: /*        Because classification of a single vector takes less than    */
wolffd@0: /*        0.01 secs, the timer was underflowing.                       */
wolffd@0:     printf("Runtime (without IO) in cpu-seconds: %.2f\n",
wolffd@0: 	   (float)(runtime/100.0));
wolffd@0:     
wolffd@0:   }
wolffd@0:   if((!no_accuracy) && (verbosity>=1)) {
wolffd@0:     printf("Accuracy on test set: %.2f%% (%ld correct, %ld incorrect, %ld total)\n",(float)(correct)*100.0/totdoc,correct,incorrect,totdoc);
wolffd@0:     printf("Precision/recall on test set: %.2f%%/%.2f%%\n",(float)(res_a)*100.0/(res_a+res_b),(float)(res_a)*100.0/(res_a+res_c));
wolffd@0:   }
wolffd@0: 
wolffd@0:   return(0);
wolffd@0: }
wolffd@0: 
wolffd@0: void read_input_parameters(int argc, char **argv, char *docfile, 
wolffd@0: 			   char *modelfile, char *predictionsfile, 
wolffd@0: 			   long int *verbosity, long int *pred_format)
wolffd@0: {
wolffd@0:   long i;
wolffd@0:   
wolffd@0:   /* set default */
wolffd@0:   strcpy (modelfile, "svm_model");
wolffd@0:   strcpy (predictionsfile, "svm_predictions"); 
wolffd@0:   (*verbosity)=2;
wolffd@0:   (*pred_format)=1;
wolffd@0: 
wolffd@0:   for(i=1;(i<argc) && ((argv[i])[0] == '-');i++) {
wolffd@0:     switch ((argv[i])[1]) 
wolffd@0:       { 
wolffd@0:       case 'h': print_help(); exit(0);
wolffd@0:       case 'v': i++; (*verbosity)=atol(argv[i]); break;
wolffd@0:       case 'f': i++; (*pred_format)=atol(argv[i]); break;
wolffd@0:       default: printf("\nUnrecognized option %s!\n\n",argv[i]);
wolffd@0: 	       print_help();
wolffd@0: 	       exit(0);
wolffd@0:       }
wolffd@0:   }
wolffd@0:   if((i+1)>=argc) {
wolffd@0:     printf("\nNot enough input parameters!\n\n");
wolffd@0:     print_help();
wolffd@0:     exit(0);
wolffd@0:   }
wolffd@0:   strcpy (docfile, argv[i]);
wolffd@0:   strcpy (modelfile, argv[i+1]);
wolffd@0:   if((i+2)<argc) {
wolffd@0:     strcpy (predictionsfile, argv[i+2]);
wolffd@0:   }
wolffd@0:   if(((*pred_format) != 0) && ((*pred_format) != 1)) {
wolffd@0:     printf("\nOutput format can only take the values 0 or 1!\n\n");
wolffd@0:     print_help();
wolffd@0:     exit(0);
wolffd@0:   }
wolffd@0: }
wolffd@0: 
wolffd@0: void print_help(void)
wolffd@0: {
wolffd@0:   printf("\nSVM-light %s: Support Vector Machine, classification module     %s\n",VERSION,VERSION_DATE);
wolffd@0:   copyright_notice();
wolffd@0:   printf("   usage: svm_classify [options] example_file model_file output_file\n\n");
wolffd@0:   printf("options: -h         -> this help\n");
wolffd@0:   printf("         -v [0..3]  -> verbosity level (default 2)\n");
wolffd@0:   printf("         -f [0,1]   -> 0: old output format of V1.0\n");
wolffd@0:   printf("                    -> 1: output the value of decision function (default)\n\n");
wolffd@0: }
wolffd@0: 
wolffd@0: 
wolffd@0: 
wolffd@0: