Changeset: d2816518acc4 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=d2816518acc4
Modified Files:
monetdb5/extras/rdf/rdflabels.c
monetdb5/extras/rdf/rdfparams.c
monetdb5/extras/rdf/rdfparams.h
monetdb5/extras/rdf/rdfschema.c
monetdb5/extras/rdf/rdfschema.h
Branch: rdf
Log Message:
Add params
diffs (106 lines):
diff --git a/monetdb5/extras/rdf/rdflabels.c b/monetdb5/extras/rdf/rdflabels.c
--- a/monetdb5/extras/rdf/rdflabels.c
+++ b/monetdb5/extras/rdf/rdflabels.c
@@ -23,6 +23,7 @@
#include "rdfschema.h"
#include "tokenizer.h"
#include <math.h>
+#include "rdfparams.h"
// list of known ontologies
int ontologyCount = 74;
@@ -830,7 +831,7 @@ oid* getOntologyCandidates(oid** ontattr
for (k = 0; k < num; ++k) {
int found = 0;
//if (freqId == 161) printf(" TFIDF score at %d
("BUNFMT") is: %f | Number of matched Prop %d \n",k, classStat[k].ontoClass,
classStat[k].tfidfs,classStat[k].numMatchedProp);
- if (classStat[k].tfidfs < SIM_TFIDF_THRESHOLD) break;
// values not frequent enough (list is sorted by tfidfs)
+ if (classStat[k].tfidfs < simTfidfThreshold) break; //
values not frequent enough (list is sorted by tfidfs)
for (j = 0; j < ontmetadataCount && (found == 0); ++j) {
oid muri = ontmetadata[0][j];
oid msuper = ontmetadata[1][j];
diff --git a/monetdb5/extras/rdf/rdfparams.c b/monetdb5/extras/rdf/rdfparams.c
--- a/monetdb5/extras/rdf/rdfparams.c
+++ b/monetdb5/extras/rdf/rdfparams.c
@@ -28,9 +28,9 @@
#include <string.h>
int dimensionFactor;
-float ontologySimThreshold;
int upperboundNumTables;
float generalityThreshold;
+float simTfidfThreshold;
void createDefaultParamsFile(void){
@@ -38,9 +38,9 @@ void createDefaultParamsFile(void){
paramFile = fopen("params.ini", "wt");
- fprintf(paramFile, "dimensionFactor 3\n");
- fprintf(paramFile, "ontologySimThreshold 0.75\n");
+ fprintf(paramFile, "dimensionFactor 1000\n");
fprintf(paramFile, "upperboundNumTables 1000");
+ fprintf(paramFile, "simTfidfThreshold 0.75");
fclose(paramFile);
}
@@ -63,14 +63,14 @@ void readParamsInput(void){
dimensionFactor = atoi(value);
printf("dimensionFactor =
%d\n",dimensionFactor);
}
- else if (strcmp(variable, "ontologySimThreshold") == 0){
- ontologySimThreshold = atof(value);
- printf("ontologySimThreshold =
%f\n",ontologySimThreshold);
- }
else if (strcmp(variable, "upperboundNumTables") == 0){
upperboundNumTables = atoi(value);
printf("upperboundNumTables = %d\n",
upperboundNumTables);
}
+ else if (strcmp(variable, "simTfidfThreshold") == 0){
+ simTfidfThreshold = atof(value);
+ printf("simTfidfThreshold = %f\n",
simTfidfThreshold);
+ }
}
}
diff --git a/monetdb5/extras/rdf/rdfparams.h b/monetdb5/extras/rdf/rdfparams.h
--- a/monetdb5/extras/rdf/rdfparams.h
+++ b/monetdb5/extras/rdf/rdfparams.h
@@ -32,9 +32,9 @@
extern int dimensionFactor;
-extern float ontologySimThreshold;
extern int upperboundNumTables;
extern float generalityThreshold;
+extern float simTfidfThreshold;
rdf_export void
createDefaultParamsFile(void);
diff --git a/monetdb5/extras/rdf/rdfschema.c b/monetdb5/extras/rdf/rdfschema.c
--- a/monetdb5/extras/rdf/rdfschema.c
+++ b/monetdb5/extras/rdf/rdfschema.c
@@ -4683,9 +4683,9 @@ void mergeCSByS4(CSset *freqCSset, CSlab
isSameLabel = 0;
if ((*labels)[freqId1].name ==
(*labels)[freqId2].name) isSameLabel = 1;
- if (simscore > SIM_TFIDF_THRESHOLD &&
(existDiscriminatingProp || isSameLabel)){
+ if (simscore > simTfidfThreshold &&
(existDiscriminatingProp || isSameLabel)){
#else
- if (simscore > SIM_TFIDF_THRESHOLD &&
existDiscriminatingProp){
+ if (simscore > simTfidfThreshold &&
existDiscriminatingProp){
#endif
#else
if (simscore > SIM_THRESHOLD) {
diff --git a/monetdb5/extras/rdf/rdfschema.h b/monetdb5/extras/rdf/rdfschema.h
--- a/monetdb5/extras/rdf/rdfschema.h
+++ b/monetdb5/extras/rdf/rdfschema.h
@@ -234,7 +234,7 @@ typedef struct SubCSSet{
#define INIT_NUM_CS 1000
#define SIM_THRESHOLD 0.6
-#define SIM_TFIDF_THRESHOLD 0.75
+//#define SIM_TFIDF_THRESHOLD 0.75 //Replaced by simTfidfThreshold
//#define IMPORTANCE_THRESHOLD 0.001 //This is used when merging CS's by
common ancestor
// Replace by generalityThreshold =
1/(upperboundNumTables)
#define COMMON_ANCESTOR_LOWEST_SPECIFIC_LEVEL 2
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list