Changeset: efb882d6b146 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=efb882d6b146
Modified Files:
monetdb5/extras/rdf/rdf.h
monetdb5/extras/rdf/rdflabels.c
monetdb5/extras/rdf/rdfschema.h
Branch: rdf
Log Message:
Merge with Linnea's changes for updatelabel function
diffs (truncated from 337 to 300 lines):
diff --git a/monetdb5/extras/rdf/rdf.h b/monetdb5/extras/rdf/rdf.h
--- a/monetdb5/extras/rdf/rdf.h
+++ b/monetdb5/extras/rdf/rdf.h
@@ -108,6 +108,10 @@ typedef struct CSlabel {
oid name; // table name
oid *candidates; // list of table name candidates,
candidates[0] == name
int candidatesCount;// number of entries in the candidates
list
+ int candidatesNew; // number of candidates that
are created during merging (e.g. ancestor name)
+ int candidatesOntology; // number of ontology
candidates (first category)
+ int candidatesType; // number of type candidates
(second category)
+ int candidatesFK; // number of fk candidates
(third category)
oid *hierarchy; // hierarchy "bottom to top"
int hierarchyCount; // number of entries in the hierarchy
list
int numProp; // number of properties, copied from
freqCSset->items[x].numProp
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
@@ -1830,6 +1830,50 @@ oid* getOntoHierarchy(oid ontology, int*
return hierarchy;
}
+/* Remove duplicated candidate values and remove DUMMY values if better
candidates exist
+ */
+static
+void removeDuplicatedCandidates(CSlabel *label) {
+ int i, j;
+ int cNew = label->candidatesNew, cOnto = label->candidatesOntology,
cType = label->candidatesType, cFK = label->candidatesFK;
+
+ if (label->candidatesCount < 2) return; // no duplicates
+
+ // loop through all candidates
+ for (i = 0; i < label->candidatesCount - 1; ++i) {
+ // search (direction: right) whether this value occurs again
+ int moveLeft = 0;
+ for (j = i + 1; j < label->candidatesCount; ++j) {
+ // find out which category (new, onto, type, fk) we are
in
+ int *cPtr = NULL;
+ if (j < label->candidatesNew) cPtr = &cNew;
+ else if (j < label->candidatesNew +
label->candidatesOntology) cPtr = &cOnto;
+ else if (j < label->candidatesNew +
label->candidatesOntology + label->candidatesType) cPtr = &cType;
+ else cPtr = &cFK;
+
+ if (label->candidates[i] == label->candidates[j] ||
label->candidates[j] == BUN_NONE) {
+ // DUMMY value will be overwritten
+ // OR:
+ // value occurs again, will be overwritten
+ moveLeft++;
+ (*cPtr)--;
+ } else {
+ // different value, keep it
+ label->candidates[j - moveLeft] =
label->candidates[j];
+ }
+ }
+ // value 'i' is unique now
+ // update counts
+ label->candidatesCount -= moveLeft;
+ label->candidatesNew = cNew;
+ label->candidatesOntology = cOnto;
+ label->candidatesType = cType;
+ label->candidatesFK = cFK;
+ }
+
+ // DUMMY value on position 0 is kept to ensure that name ==
candidates[0]
+}
+
#if USE_TABLE_NAME
/* For one CS: Choose the best table name out of all collected candidates
(ontology, type, fk). */
static
@@ -1842,6 +1886,7 @@ void getTableName(CSlabel* label, int cs
// --- ONTOLOGY ---
// add all ontology candidates to list of candidates
if (resultCount[csIdx] >= 1) {
+ label->candidatesOntology = resultCount[csIdx];
label->candidates = GDKrealloc(label->candidates, sizeof(oid) *
(label->candidatesCount + resultCount[csIdx]));
if (!label->candidates) fprintf(stderr, "ERROR: Couldn't
realloc memory!\n");
for (i = 0; i < resultCount[csIdx]; ++i) {
@@ -1924,6 +1969,7 @@ void getTableName(CSlabel* label, int cs
// add all most frequent type values to list of candidates
if (tmpListCount >= 1) {
int counter = 0;
+ label->candidatesType = tmpListCount;
label->candidates = GDKrealloc(label->candidates, sizeof(oid) *
(label->candidatesCount + tmpListCount));
if (!label->candidates) fprintf(stderr, "ERROR: Couldn't
realloc memory!\n");
for (i = 0; i < typeStatCount; ++i) {
@@ -1964,6 +2010,7 @@ void getTableName(CSlabel* label, int cs
// --- FK ---
// add top3 fk values to list of candidates
if (links[csIdx].num > 0) {
+ label->candidatesFK = MIN(3, links[csIdx].num);
label->candidates = GDKrealloc(label->candidates, sizeof(oid) *
(label->candidatesCount + MIN(3, links[csIdx].num)));
if (!label->candidates) fprintf(stderr, "ERROR: Couldn't
realloc memory!\n");
for (i = 0; i < MIN(3, links[csIdx].num); ++i) {
@@ -1982,6 +2029,7 @@ void getTableName(CSlabel* label, int cs
// --- NOTHING ---
if (label->candidatesCount == 0) {
+ label->candidatesNew = 1;
label->candidates = GDKrealloc(label->candidates, sizeof(oid));
if (!label->candidates) fprintf(stderr, "ERROR: Couldn't
realloc memory!\n");
label->candidates[0] = BUN_NONE;
@@ -1993,6 +2041,9 @@ void getTableName(CSlabel* label, int cs
nameFound = 1;
}
+ // de-duplicate
+ removeDuplicatedCandidates(label);
+
if(tmpList != NULL) free(tmpList);
return;
}
@@ -2009,6 +2060,10 @@ CSlabel* initLabels(CSset *freqCSset) {
labels[i].name = BUN_NONE;
labels[i].candidates = NULL;
labels[i].candidatesCount = 0;
+ labels[i].candidatesNew = 0;
+ labels[i].candidatesOntology = 0;
+ labels[i].candidatesType = 0;
+ labels[i].candidatesFK = 0;
labels[i].hierarchy = NULL;
labels[i].hierarchyCount = 0;
labels[i].numProp = 0;
@@ -2450,6 +2505,68 @@ CSlabel* createLabels(CSset* freqCSset,
return labels;
}
+/* Merge two lists of candidates.
+ * Result: <common name> <ontology candidates CS1> <ontology candidates CS2>
<type candidates CS1> <type candidates CS2> <FK candidates CS1> <FK candidates
CS2>
+ */
+static
+oid* mergeCandidates(int *candidatesCount, int *candidatesNew, int
*candidatesOntology, int *candidatesType, int *candidatesFK, CSlabel cs1,
CSlabel cs2, oid commonName) {
+ oid *candidates;
+ int counter = 0;
+ int i;
+
+ (*candidatesCount) = cs1.candidatesCount + cs2.candidatesCount + 1; //
+1 for common name
+ candidates = GDKmalloc(sizeof(oid) * (*candidatesCount));
+
+ candidates[counter] = commonName;
+ counter++;
+
+ // copy "new"
+ for (i = 0; i < cs1.candidatesNew; ++i) {
+ candidates[counter] = cs1.candidates[i];
+ counter++;
+ }
+ for (i = 0; i < cs2.candidatesNew; ++i) {
+ candidates[counter] = cs2.candidates[i];
+ counter++;
+ }
+ (*candidatesNew) = counter;
+
+ // copy "ontology"
+ for (i = 0; i < cs1.candidatesOntology; ++i) {
+ candidates[counter] = cs1.candidates[cs1.candidatesNew + i];
+ counter++;
+ }
+ for (i = 0; i < cs2.candidatesOntology; ++i) {
+ candidates[counter] = cs2.candidates[cs2.candidatesNew + i];
+ counter++;
+ }
+ (*candidatesOntology) = counter - (*candidatesNew);
+
+ // copy "type"
+ for (i = 0; i < cs1.candidatesType; ++i) {
+ candidates[counter] = cs1.candidates[cs1.candidatesNew +
cs1.candidatesOntology + i];
+ counter++;
+ }
+ for (i = 0; i < cs2.candidatesType; ++i) {
+ candidates[counter] = cs2.candidates[cs2.candidatesNew +
cs2.candidatesOntology + i];
+ counter++;
+ }
+ (*candidatesType) = counter - (*candidatesNew) - (*candidatesOntology);
+
+ // copy "fk"
+ for (i = 0; i < cs1.candidatesFK; ++i) {
+ candidates[counter] = cs1.candidates[cs1.candidatesNew +
cs1.candidatesOntology + cs1.candidatesType + i];
+ counter++;
+ }
+ for (i = 0; i < cs2.candidatesFK; ++i) {
+ candidates[counter] = cs2.candidates[cs2.candidatesNew +
cs2.candidatesOntology + cs2.candidatesType + i];
+ counter++;
+ }
+ (*candidatesFK) = counter - (*candidatesNew) - (*candidatesOntology) -
(*candidatesType);
+
+ return candidates;
+}
+
/* Create labels for merged CS's. Uses rules S1 to S5 (new names!).
* If no MERGECS is created (subset-superset relation), mergeCSFreqId contains
the Id of the superset class.
* For S1 and S2, parameter 'name' is used to avoid recomputation of CS names
@@ -2457,15 +2574,19 @@ CSlabel* createLabels(CSset* freqCSset,
str updateLabel(int ruleNumber, CSset *freqCSset, CSlabel **labels, int newCS,
int mergeCSFreqId, int freqCS1, int freqCS2, oid name, oid **ontmetadata, int
ontmetadataCount, int *lstFreqId, int numIds){
int i;
int freqCS1Counter;
- CSlabel *big;
+ CSlabel big, small;
CSlabel *label;
CS cs;
#if USE_MULTIWAY_MERGING
int tmpMaxCoverage;
int tmpFreqId;
#endif
+ oid *mergedCandidates = NULL;
+ int candidatesCount, candidatesNew, candidatesOntology,
candidatesType, candidatesFK;
+
(void) lstFreqId;
(void) numIds;
+
if (newCS) {
// realloc labels
*labels = GDKrealloc(*labels, sizeof(CSlabel) *
freqCSset->numCSadded);
@@ -2473,6 +2594,10 @@ str updateLabel(int ruleNumber, CSset *f
(*labels)[mergeCSFreqId].name = BUN_NONE;
(*labels)[mergeCSFreqId].candidates = NULL;
(*labels)[mergeCSFreqId].candidatesCount = 0;
+ (*labels)[mergeCSFreqId].candidatesNew = 0;
+ (*labels)[mergeCSFreqId].candidatesOntology = 0;
+ (*labels)[mergeCSFreqId].candidatesType = 0;
+ (*labels)[mergeCSFreqId].candidatesFK = 0;
(*labels)[mergeCSFreqId].hierarchy = NULL;
(*labels)[mergeCSFreqId].hierarchyCount = 0;
(*labels)[mergeCSFreqId].numProp = 0;
@@ -2503,9 +2628,16 @@ str updateLabel(int ruleNumber, CSset *f
(void)freqCS2;
#else
- // TODO candidates
- //label->candidates = ;
- //label->candidatesCount = ;
+ // candidates
+ mergedCandidates = mergeCandidates(&candidatesCount,
&candidatesNew, &candidatesOntology, &candidatesType, &candidatesFK,
(*labels)[freqCS1], (*labels)[freqCS2], label->name);
+ GDKfree(label->candidates);
+ label->candidates = mergedCandidates; // TODO check access
outside function
+ label->candidatesCount = candidatesCount;
+ label->candidatesNew = candidatesNew;
+ label->candidatesOntology = candidatesOntology;
+ label->candidatesType = candidatesType;
+ label->candidatesFK = candidatesFK;
+ removeDuplicatedCandidates(label);
// hierarchy
if ((*labels)[freqCS1].name == label->name) {
@@ -2541,9 +2673,16 @@ str updateLabel(int ruleNumber, CSset *f
// use common ancestor
label->name = name;
- // TODO candidates
- //label->candidates = ;
- //label->candidatesCount = ;
+ // candidates
+ mergedCandidates = mergeCandidates(&candidatesCount,
&candidatesNew, &candidatesOntology, &candidatesType, &candidatesFK,
(*labels)[freqCS1], (*labels)[freqCS2], label->name);
+ GDKfree(label->candidates);
+ label->candidates = mergedCandidates; // TODO check access
outside function
+ label->candidatesCount = candidatesCount;
+ label->candidatesNew = candidatesNew;
+ label->candidatesOntology = candidatesOntology;
+ label->candidatesType = candidatesType;
+ label->candidatesFK = candidatesFK;
+ removeDuplicatedCandidates(label);
// hierarchy
freqCS1Counter = (*labels)[freqCS1].hierarchyCount - 1;
@@ -2565,7 +2704,18 @@ str updateLabel(int ruleNumber, CSset *f
case S3:
// subset-superset relation
- // candidates already set
+
+ // candidates
+ mergedCandidates = mergeCandidates(&candidatesCount,
&candidatesNew, &candidatesOntology, &candidatesType, &candidatesFK,
(*labels)[freqCS1], (*labels)[freqCS2], label->name); // freqCS1 is superCS,
freqCS2 is subCS
+ GDKfree(label->candidates);
+ label->candidates = mergedCandidates; // TODO check access
outside function
+ label->candidatesCount = candidatesCount;
+ label->candidatesNew = candidatesNew;
+ label->candidatesOntology = candidatesOntology;
+ label->candidatesType = candidatesType;
+ label->candidatesFK = candidatesFK;
+ removeDuplicatedCandidates(label);
+
// hierarchy already set
// properties already set
@@ -2587,25 +2737,34 @@ str updateLabel(int ruleNumber, CSset *f
#else
// use label of biggest CS (higher coverage value)
if (freqCSset->items[freqCS1].coverage >
freqCSset->items[freqCS2].coverage) {
- big = &(*labels)[freqCS1];
+ big = (*labels)[freqCS1];
+ small = (*labels)[freqCS2];
} else {
- big = &(*labels)[freqCS2];
+ big = (*labels)[freqCS2];
+ small = (*labels)[freqCS1];
}
#endif
- label->name = big->name;
+ label->name = big.name;
- // TODO candidates
- //label->candidatesCount = ;
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
