Menu
▾
▴
[Bigdata-commit] SF.net SVN: bigdata:[4034] branches/JOURNAL_HA_BRANCH/bigdata-sails/src
|
From: <mrp...@us...> - 2010-12-21 19:30:57
|
Revision: 4034
http://bigdata.svn.sourceforge.net/bigdata/?rev=4034&view=rev
Author: mrpersonick
Date: 2010-12-21 19:30:51 +0000 (Tue, 21 Dec 2010)
Log Message:
-----------
added support for queries specifying text search maxHits and minRelevance
Modified Paths:
--------------
branches/JOURNAL_HA_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/BigdataEvaluationStrategyImpl2.java
branches/JOURNAL_HA_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/FreeTextSearchExpander.java
branches/JOURNAL_HA_BRANCH/bigdata-sails/src/test/com/bigdata/rdf/sail/TestSearchQuery.java
Modified: branches/JOURNAL_HA_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/BigdataEvaluationStrategyImpl2.java
===================================================================
--- branches/JOURNAL_HA_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/BigdataEvaluationStrategyImpl2.java 2010-12-21 16:25:29 UTC (rev 4033)
+++ branches/JOURNAL_HA_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/BigdataEvaluationStrategyImpl2.java 2010-12-21 19:30:51 UTC (rev 4034)
@@ -794,9 +794,10 @@
final StatementPattern sp = it.next().getKey();
final Value s = sp.getSubjectVar().getValue();
final Value p = sp.getPredicateVar().getValue();
- final Value o = sp.getObjectVar().getValue();
- if (s == null && p != null && o == null) {
- if (BD.RELEVANCE.equals(p)) {
+ if (s == null && p != null) {
+ if (BD.RELEVANCE.equals(p) ||
+ BD.MAX_HITS.equals(p) ||
+ BD.MIN_RELEVANCE.equals(p)) {
final Var sVar = sp.getSubjectVar();
Set<StatementPattern> metadata = searchMetadata2.get(sVar);
if (metadata != null) {
@@ -1507,15 +1508,14 @@
throw new IllegalArgumentException("not a valid magic search: " + sp);
}
- final ISolutionExpander expander =
- new FreeTextSearchExpander(database, (Literal) objValue);
-
final Var subjVar = sp.getSubjectVar();
final IVariableOrConstant<IV> search =
com.bigdata.relation.rule.Var.var(subjVar.getName());
IVariableOrConstant<IV> relevance = new Constant(DummyIV.INSTANCE);
+ Literal maxHits = null;
+ Literal minRelevance = null;
for (StatementPattern meta : metadata) {
if (!meta.getSubjectVar().equals(subjVar)) {
@@ -1523,14 +1523,32 @@
}
final Value pVal = meta.getPredicateVar().getValue();
final Var oVar = meta.getObjectVar();
- if (pVal == null || oVar.hasValue()) {
+ final Value oVal = oVar.getValue();
+ if (pVal == null) {
throw new IllegalArgumentException("illegal metadata: " + meta);
}
if (BD.RELEVANCE.equals(pVal)) {
+ if (oVar.hasValue()) {
+ throw new IllegalArgumentException("illegal metadata: " + meta);
+ }
relevance = com.bigdata.relation.rule.Var.var(oVar.getName());
- }
+ } else if (BD.MAX_HITS.equals(pVal)) {
+ if (oVal == null || !(oVal instanceof Literal)) {
+ throw new IllegalArgumentException("illegal metadata: " + meta);
+ }
+ maxHits = (Literal) oVal;
+ } else if (BD.MIN_RELEVANCE.equals(pVal)) {
+ if (oVal == null || !(oVal instanceof Literal)) {
+ throw new IllegalArgumentException("illegal metadata: " + meta);
+ }
+ minRelevance = (Literal) oVal;
+ }
}
+ final ISolutionExpander expander =
+ new FreeTextSearchExpander(database, (Literal) objValue,
+ maxHits, minRelevance);
+
return new SPOPredicate(
new String[] { database.getSPORelation().getNamespace() },
-1, // partitionId
Modified: branches/JOURNAL_HA_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/FreeTextSearchExpander.java
===================================================================
--- branches/JOURNAL_HA_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/FreeTextSearchExpander.java 2010-12-21 16:25:29 UTC (rev 4033)
+++ branches/JOURNAL_HA_BRANCH/bigdata-sails/src/java/com/bigdata/rdf/sail/FreeTextSearchExpander.java 2010-12-21 19:30:51 UTC (rev 4034)
@@ -56,12 +56,18 @@
private final AbstractTripleStore database;
- private final Literal query;
+ private final Literal query, maxHits, minRelevance;
private Set<URI> graphs;
public FreeTextSearchExpander(final AbstractTripleStore database,
final Literal query) {
+ this(database, query, null, null);
+ }
+
+ public FreeTextSearchExpander(final AbstractTripleStore database,
+ final Literal query, final Literal maxHits,
+ final Literal minRelevance) {
if (database == null)
throw new IllegalArgumentException();
@@ -73,6 +79,10 @@
this.query = query;
+ this.maxHits = maxHits;
+
+ this.minRelevance = minRelevance;
+
}
public boolean backchain() {
@@ -135,8 +145,10 @@
// final long begin = System.nanoTime();
hiterator = database.getLexiconRelation()
.getSearchEngine().search(query.getLabel(),
- query.getLanguage(), false/* prefixMatch */,
- 0d/* minCosine */, 10000/* maxRank */,
+ query.getLanguage(),
+ false/* prefixMatch */,
+ minRelevance == null ? 0d : minRelevance.doubleValue()/* minCosine */,
+ maxHits == null ? 10000 : maxHits.intValue()+1/* maxRank */,
1000L/* timeout */, TimeUnit.MILLISECONDS);
// hiterator = database.getSearchEngine().search
// ( query.getLabel(),
Modified: branches/JOURNAL_HA_BRANCH/bigdata-sails/src/test/com/bigdata/rdf/sail/TestSearchQuery.java
===================================================================
--- branches/JOURNAL_HA_BRANCH/bigdata-sails/src/test/com/bigdata/rdf/sail/TestSearchQuery.java 2010-12-21 16:25:29 UTC (rev 4033)
+++ branches/JOURNAL_HA_BRANCH/bigdata-sails/src/test/com/bigdata/rdf/sail/TestSearchQuery.java 2010-12-21 19:30:51 UTC (rev 4034)
@@ -677,7 +677,7 @@
}
- public void testWithRelevance() throws Exception {
+ public void testWithMetadata() throws Exception {
final BigdataSail sail = getSail();
try {
@@ -755,16 +755,19 @@
" ?s <"+RDFS.LABEL+"> ?o . " +
" ?o <"+BD.SEARCH+"> \""+searchQuery+"\" . " +
" ?o <"+BD.RELEVANCE+"> ?score . " +
- "}";
+ "} " +
+ "order by desc(?score)";
final TupleQuery tupleQuery =
cxn.prepareTupleQuery(QueryLanguage.SPARQL, query);
tupleQuery.setIncludeInferred(true /* includeInferred */);
TupleQueryResult result = tupleQuery.evaluate();
+ int i = 0;
while (result.hasNext()) {
- System.err.println(result.next());
+ System.err.println(i++ + ": " + result.next().toString());
}
+ assertTrue("wrong # of results", i == 7);
result = tupleQuery.evaluate();
@@ -777,7 +780,7 @@
null, // languageCode
false, // prefixMatch
0d, // minCosine
- 10000, // maxRank
+ 10000, // maxRank (=maxResults + 1)
1000L, // timeout
TimeUnit.MILLISECONDS // unit
);
@@ -788,16 +791,140 @@
final Literal score = vf.createLiteral(hit.getCosine());
final URI s = uris.get(id);
final Literal o = literals.get(id);
- answer.add(createBindingSet(
+ final BindingSet bs = createBindingSet(
new BindingImpl("s", s),
new BindingImpl("o", o),
- new BindingImpl("score", score)));
+ new BindingImpl("score", score));
+ System.err.println(bs);
+ answer.add(bs);
}
compare(result, answer);
}
+ {
+ final String searchQuery = "how now brown cow";
+ final int maxHits = 5;
+
+ final String query =
+ "select ?s ?o ?score " +
+ "where " +
+ "{ " +
+ " ?s <"+RDFS.LABEL+"> ?o . " +
+ " ?o <"+BD.SEARCH+"> \""+searchQuery+"\" . " +
+ " ?o <"+BD.RELEVANCE+"> ?score . " +
+// " ?o <"+BD.MIN_RELEVANCE+"> \"0.6\" . " +
+ " ?o <"+BD.MAX_HITS+"> \""+maxHits+"\" . " +
+ "} " +
+ "order by desc(?score)";
+
+ final TupleQuery tupleQuery =
+ cxn.prepareTupleQuery(QueryLanguage.SPARQL, query);
+ tupleQuery.setIncludeInferred(true /* includeInferred */);
+ TupleQueryResult result = tupleQuery.evaluate();
+
+ int i = 0;
+ while (result.hasNext()) {
+ System.err.println(i++ + ": " + result.next().toString());
+ }
+ assertTrue("wrong # of results", i == 5);
+
+ result = tupleQuery.evaluate();
+
+ Collection<BindingSet> answer = new LinkedList<BindingSet>();
+
+ final ITextIndexer search =
+ sail.getDatabase().getLexiconRelation().getSearchEngine();
+ final Hiterator<IHit> hits =
+ search.search(searchQuery,
+ null, // languageCode
+ false, // prefixMatch
+ 0d, // minCosine
+ maxHits+1, // maxRank (=maxResults + 1)
+ 1000L, // timeout
+ TimeUnit.MILLISECONDS // unit
+ );
+
+ while (hits.hasNext()) {
+ final IHit hit = hits.next();
+ final IV id = new TermId(VTE.LITERAL, hit.getDocId());
+ final Literal score = vf.createLiteral(hit.getCosine());
+ final URI s = uris.get(id);
+ final Literal o = literals.get(id);
+ final BindingSet bs = createBindingSet(
+ new BindingImpl("s", s),
+ new BindingImpl("o", o),
+ new BindingImpl("score", score));
+ System.err.println(bs);
+ answer.add(bs);
+ }
+
+ compare(result, answer);
+
+ }
+
+ {
+ final String searchQuery = "how now brown cow";
+ final double minRelevance = 0.6d;
+
+ final String query =
+ "select ?s ?o ?score " +
+ "where " +
+ "{ " +
+ " ?s <"+RDFS.LABEL+"> ?o . " +
+ " ?o <"+BD.SEARCH+"> \""+searchQuery+"\" . " +
+ " ?o <"+BD.RELEVANCE+"> ?score . " +
+ " ?o <"+BD.MIN_RELEVANCE+"> \""+minRelevance+"\" . " +
+// " ?o <"+BD.MAX_HITS+"> \"5\" . " +
+ "} " +
+ "order by desc(?score)";
+
+ final TupleQuery tupleQuery =
+ cxn.prepareTupleQuery(QueryLanguage.SPARQL, query);
+ tupleQuery.setIncludeInferred(true /* includeInferred */);
+ TupleQueryResult result = tupleQuery.evaluate();
+
+ int i = 0;
+ while (result.hasNext()) {
+ System.err.println(i++ + ": " + result.next().toString());
+ }
+ assertTrue("wrong # of results", i == 3);
+
+ result = tupleQuery.evaluate();
+
+ Collection<BindingSet> answer = new LinkedList<BindingSet>();
+
+ final ITextIndexer search =
+ sail.getDatabase().getLexiconRelation().getSearchEngine();
+ final Hiterator<IHit> hits =
+ search.search(searchQuery,
+ null, // languageCode
+ false, // prefixMatch
+ minRelevance, // minCosine
+ 10000, // maxRank (=maxResults + 1)
+ 1000L, // timeout
+ TimeUnit.MILLISECONDS // unit
+ );
+
+ while (hits.hasNext()) {
+ final IHit hit = hits.next();
+ final IV id = new TermId(VTE.LITERAL, hit.getDocId());
+ final Literal score = vf.createLiteral(hit.getCosine());
+ final URI s = uris.get(id);
+ final Literal o = literals.get(id);
+ final BindingSet bs = createBindingSet(
+ new BindingImpl("s", s),
+ new BindingImpl("o", o),
+ new BindingImpl("score", score));
+ System.err.println(bs);
+ answer.add(bs);
+ }
+
+ compare(result, answer);
+
+ }
+
} finally {
cxn.close();
}
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
