Search in sources :

Example 11 with SpanRelationQuery

use of de.ids_mannheim.korap.query.SpanRelationQuery in project Krill by KorAP.

the class TestReferenceIndex method testCase2.

@Test
public void testCase2() throws IOException, QueryException {
    String filepath = getClass().getResource("/queries/reference/distance-reference.jsonld").getFile();
    SpanQueryWrapper sqwi = getJSONQuery(filepath);
    SpanQuery sq = sqwi.toQuery();
    // cat="vb" & cat="prp" & cat="nn" & #1 .{0,1} #2 & #1
    // .{0,2} #3 & #3 -> #2
    assertEquals("spanReference(focus(#[1,2]spanSegment({2: <tokens:prp />}, focus(#2: " + "spanSegment(spanRelation(tokens:>:stanford/d:tag), " + "focus(3: spanDistance(focus(1: spanDistance({1: <tokens:vb />}, " + "{2: <tokens:prp />}, [(w[1:2], notOrdered, notExcluded)])), " + "{3: <tokens:nn />}, [(w[1:3], notOrdered, notExcluded)]))))" + ")), 2)", sq.toString());
    SpanElementQuery seq1 = new SpanElementQuery("tokens", "vb");
    // new SpanTermQuery(new Term("tokens", "c:vb"));
    SpanElementQuery seq2 = new SpanElementQuery("tokens", "prp");
    // new SpanTermQuery(new Term("tokens", "c:prp"));
    SpanElementQuery seq3 = new SpanElementQuery("tokens", "nn");
    // new SpanTermQuery(new Term("tokens", "c:nn"));
    SpanClassQuery scq1 = new SpanClassQuery(seq1, (byte) 1);
    SpanClassQuery scq2 = new SpanClassQuery(seq2, (byte) 2);
    SpanClassQuery scq3 = new SpanClassQuery(seq3, (byte) 3);
    SpanDistanceQuery sdq1 = new SpanDistanceQuery(scq1, scq2, new DistanceConstraint(1, 2, false, false), true);
    SpanFocusQuery sfq1 = new SpanFocusQuery(sdq1, (byte) 1);
    SpanDistanceQuery sdq2 = new SpanDistanceQuery(sfq1, scq3, new DistanceConstraint(1, 3, false, false), true);
    SpanFocusQuery sfq2 = new SpanFocusQuery(sdq2, (byte) 3);
    // nn -> prp
    SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(new Term("tokens", ">:stanford/d:tag")), true, RelationDirection.RIGHT);
    SpanRelationMatchQuery rq = new SpanRelationMatchQuery(srq, sfq2, scq2, true);
    SpanReferenceQuery ref = new SpanReferenceQuery(rq, (byte) 2, true);
    ki = new KrillIndex();
    ki.addDoc(createFieldDoc0());
    ki.commit();
    kr = ki.search(ref, (short) 10);
    // for (Match km : kr.getMatches()) {
    // System.out.println(km.getStartPos() + "," + km.getEndPos()
    // + " "
    // + km.getSnippetBrackets());
    // }
    assertEquals(sq.toString(), ref.toString());
    assertEquals(2, kr.getMatch(0).getStartPos());
    assertEquals(4, kr.getMatch(0).getEndPos());
    assertEquals(5, kr.getMatch(1).getStartPos());
    assertEquals(9, kr.getMatch(1).getEndPos());
    assertEquals(11, kr.getMatch(2).getStartPos());
    assertEquals(13, kr.getMatch(2).getEndPos());
    // multiple references
    SpanFocusQuery sfq3 = new SpanFocusQuery(ref, (byte) 1);
    // vp -> nn
    SpanRelationMatchQuery rq2 = new SpanRelationMatchQuery(srq, sfq3, scq3, true);
    SpanReferenceQuery ref2 = new SpanReferenceQuery(rq2, (byte) 3, true);
    kr = ki.search(ref2, (short) 10);
    assertEquals(1, kr.getMatch(0).getStartPos());
    assertEquals(4, kr.getMatch(0).getEndPos());
    assertEquals(10, kr.getMatch(1).getStartPos());
    assertEquals(13, kr.getMatch(1).getEndPos());
// for (Match km : kr.getMatches()) {
// System.out.println(km.getStartPos() + "," + km.getEndPos()
// + " "
// + km.getSnippetBrackets());
// }
}
Also used : SpanRelationQuery(de.ids_mannheim.korap.query.SpanRelationQuery) DistanceConstraint(de.ids_mannheim.korap.query.DistanceConstraint) Term(org.apache.lucene.index.Term) SpanReferenceQuery(de.ids_mannheim.korap.query.SpanReferenceQuery) SpanElementQuery(de.ids_mannheim.korap.query.SpanElementQuery) KrillIndex(de.ids_mannheim.korap.KrillIndex) SpanQuery(org.apache.lucene.search.spans.SpanQuery) SpanClassQuery(de.ids_mannheim.korap.query.SpanClassQuery) SpanTermQuery(org.apache.lucene.search.spans.SpanTermQuery) SpanDistanceQuery(de.ids_mannheim.korap.query.SpanDistanceQuery) SpanRelationMatchQuery(de.ids_mannheim.korap.query.SpanRelationMatchQuery) SpanQueryWrapper(de.ids_mannheim.korap.query.wrap.SpanQueryWrapper) SpanFocusQuery(de.ids_mannheim.korap.query.SpanFocusQuery) Test(org.junit.Test)

Example 12 with SpanRelationQuery

use of de.ids_mannheim.korap.query.SpanRelationQuery in project Krill by KorAP.

the class TestRelationIndex method testRelationWithRegex.

@Test
public void testRelationWithRegex() throws IOException {
    ki.addDoc(createFieldDoc0());
    ki.addDoc(createFieldDoc3());
    ki.commit();
    SpanQuery sq;
    sq = new SpanRelationQuery(new SpanMultiTermQueryWrapper<RegexQuery>(new RegexQuery(new Term("base", ">:xip/.*"))), true, RelationDirection.RIGHT);
    kr = ki.search(sq, (short) 10);
    assertEquals((long) 7, kr.getTotalResults());
    sq = new SpanRelationQuery(new SpanMultiTermQueryWrapper<RegexQuery>(new RegexQuery(new Term("base", "<:xip/.*"))), true, RelationDirection.LEFT);
    kr = ki.search(sq, (short) 10);
    assertEquals((long) 7, kr.getTotalResults());
}
Also used : SpanMultiTermQueryWrapper(org.apache.lucene.search.spans.SpanMultiTermQueryWrapper) SpanRelationQuery(de.ids_mannheim.korap.query.SpanRelationQuery) RegexQuery(org.apache.lucene.sandbox.queries.regex.RegexQuery) Term(org.apache.lucene.index.Term) SpanQuery(org.apache.lucene.search.spans.SpanQuery) Test(org.junit.Test)

Example 13 with SpanRelationQuery

use of de.ids_mannheim.korap.query.SpanRelationQuery in project Krill by KorAP.

the class TestRelationIndex method testCase6.

/**
 * Relation directions <br/>
 * Relation with specific sources, return the sources
 */
@Test
public void testCase6() throws IOException {
    ki.addDoc(createFieldDoc2());
    ki.commit();
    // return all children that are NP
    SpanElementQuery seq1 = new SpanElementQuery("base", "np");
    SpanClassQuery scq1 = new SpanClassQuery(seq1, (byte) 1);
    SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(new Term("base", ">:child-of")), true, RelationDirection.RIGHT);
    SpanRelationMatchQuery rm = new SpanRelationMatchQuery(srq, scq1, true);
    SpanFocusQuery rv = new SpanFocusQuery(rm, (byte) 1);
    rv.setSorted(false);
    assertEquals("focus(1: focus(#[1,2]spanSegment(spanRelation(base:>:child-of), {1: <base:np />})))", rv.toString());
    kr = ki.search(rv, (short) 10);
    assertEquals(4, kr.getTotalResults());
    assertEquals(0, kr.getMatch(0).getStartPos());
    assertEquals(1, kr.getMatch(0).getEndPos());
    assertEquals(2, kr.getMatch(1).getStartPos());
    assertEquals(4, kr.getMatch(1).getEndPos());
    assertEquals(2, kr.getMatch(2).getStartPos());
    assertEquals(7, kr.getMatch(2).getEndPos());
    assertEquals(5, kr.getMatch(3).getStartPos());
    assertEquals(7, kr.getMatch(3).getEndPos());
    // return all parents that are NP
    srq = new SpanRelationQuery(new SpanTermQuery(new Term("base", "<:child-of")), true, RelationDirection.LEFT);
    rm = new SpanRelationMatchQuery(srq, scq1, true);
    rv = new SpanFocusQuery(rm, (byte) 1);
    kr = ki.search(rv, (short) 10);
    assertEquals(7, kr.getTotalResults());
    assertEquals(0, kr.getMatch(0).getStartPos());
    assertEquals(1, kr.getMatch(0).getEndPos());
    assertEquals(2, kr.getMatch(1).getStartPos());
    assertEquals(4, kr.getMatch(1).getEndPos());
    assertEquals(2, kr.getMatch(2).getStartPos());
    assertEquals(4, kr.getMatch(2).getEndPos());
    assertEquals(2, kr.getMatch(3).getStartPos());
    assertEquals(7, kr.getMatch(3).getEndPos());
    assertEquals(2, kr.getMatch(4).getStartPos());
    assertEquals(7, kr.getMatch(4).getEndPos());
    assertEquals(5, kr.getMatch(5).getStartPos());
    assertEquals(7, kr.getMatch(5).getEndPos());
    assertEquals(5, kr.getMatch(6).getStartPos());
    assertEquals(7, kr.getMatch(6).getEndPos());
}
Also used : SpanClassQuery(de.ids_mannheim.korap.query.SpanClassQuery) SpanTermQuery(org.apache.lucene.search.spans.SpanTermQuery) SpanRelationQuery(de.ids_mannheim.korap.query.SpanRelationQuery) SpanRelationMatchQuery(de.ids_mannheim.korap.query.SpanRelationMatchQuery) Term(org.apache.lucene.index.Term) SpanElementQuery(de.ids_mannheim.korap.query.SpanElementQuery) SpanFocusQuery(de.ids_mannheim.korap.query.SpanFocusQuery) Test(org.junit.Test)

Example 14 with SpanRelationQuery

use of de.ids_mannheim.korap.query.SpanRelationQuery in project Krill by KorAP.

the class TestRelationIndex method testCase9b.

@Test
public void testCase9b() throws IOException {
    ki.addDoc(createFieldDoc2());
    ki.commit();
    SpanTermWithIdQuery tiq = new SpanTermWithIdQuery(new Term("base", "pos:ART"), true);
    SpanClassQuery scq1 = new SpanClassQuery(tiq, (byte) 1);
    SpanElementQuery seq = new SpanElementQuery("base", "np");
    SpanClassQuery scq2 = new SpanClassQuery(seq, (byte) 2);
    SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(new Term("base", ">:child-of")), true, RelationDirection.RIGHT);
    srq.setSourceClass((byte) 1);
    srq.setTargetClass((byte) 2);
    SpanRelationMatchQuery rm = new SpanRelationMatchQuery(srq, scq1, scq2, true);
    // return all nps whose children are articles
    SpanFocusQuery rv = new SpanFocusQuery(rm, (byte) 2);
    rv.setSorted(false);
    kr = ki.search(rv, (short) 10);
    assertEquals((long) 2, kr.getTotalResults());
    assertEquals(2, kr.getMatch(0).getStartPos());
    assertEquals(4, kr.getMatch(0).getEndPos());
    assertEquals(5, kr.getMatch(1).getStartPos());
    assertEquals(7, kr.getMatch(1).getEndPos());
// for (Match m : kr.getMatches()) {
// System.out.println(m.getStartPos() + " " + m.getEndPos());
// }
}
Also used : SpanTermWithIdQuery(de.ids_mannheim.korap.query.SpanTermWithIdQuery) SpanClassQuery(de.ids_mannheim.korap.query.SpanClassQuery) SpanTermQuery(org.apache.lucene.search.spans.SpanTermQuery) SpanRelationQuery(de.ids_mannheim.korap.query.SpanRelationQuery) SpanRelationMatchQuery(de.ids_mannheim.korap.query.SpanRelationMatchQuery) Term(org.apache.lucene.index.Term) SpanElementQuery(de.ids_mannheim.korap.query.SpanElementQuery) SpanFocusQuery(de.ids_mannheim.korap.query.SpanFocusQuery) Test(org.junit.Test)

Example 15 with SpanRelationQuery

use of de.ids_mannheim.korap.query.SpanRelationQuery in project Krill by KorAP.

the class TestRelationIndex method testCase9c.

@Test
public void testCase9c() throws IOException {
    ki.addDoc(createFieldDoc2());
    ki.commit();
    SpanTermWithIdQuery tiq = new SpanTermWithIdQuery(new Term("base", "pos:ART"), true);
    SpanClassQuery scq1 = new SpanClassQuery(tiq, (byte) 1);
    kr = ki.search(scq1, (short) 10);
    assertEquals((long) 2, kr.getTotalResults());
    assertEquals(2, kr.getMatch(0).getStartPos());
    assertEquals(3, kr.getMatch(0).getEndPos());
    assertEquals(5, kr.getMatch(1).getStartPos());
    assertEquals(6, kr.getMatch(1).getEndPos());
    SpanRelationQuery srq = new SpanRelationQuery(new SpanTermQuery(new Term("base", ">:child-of")), true, RelationDirection.RIGHT);
    srq.setSourceClass((byte) 1);
    srq.setTargetClass((byte) 2);
    // match articles as relation sources (left side)
    SpanSegmentQuery ssq1 = new SpanSegmentQuery(srq, scq1);
    // NOTE: SegmentSpans can only always adopt the relation left
    // side id.
    kr = ki.search(ssq1, (short) 10);
    assertEquals(2, kr.getMatch(0).getStartPos());
    assertEquals(3, kr.getMatch(0).getEndPos());
    assertEquals(5, kr.getMatch(1).getStartPos());
    assertEquals(6, kr.getMatch(1).getEndPos());
    // return all parents of articles
    SpanFocusQuery sfq = new SpanFocusQuery(ssq1, (byte) 2);
    sfq.setSorted(false);
    kr = ki.search(sfq, (short) 10);
    assertEquals((long) 2, kr.getTotalResults());
    assertEquals(2, kr.getMatch(0).getStartPos());
    assertEquals(4, kr.getMatch(0).getEndPos());
    assertEquals(5, kr.getMatch(1).getStartPos());
    assertEquals(7, kr.getMatch(1).getEndPos());
    SpanElementQuery seq = new SpanElementQuery("base", "np");
    SpanClassQuery scq2 = new SpanClassQuery(seq, (byte) 2);
    kr = ki.search(scq2, (short) 10);
    assertEquals((long) 4, kr.getTotalResults());
    assertEquals(0, kr.getMatch(0).getStartPos());
    assertEquals(1, kr.getMatch(0).getEndPos());
    assertEquals(2, kr.getMatch(1).getStartPos());
    assertEquals(4, kr.getMatch(1).getEndPos());
    assertEquals(2, kr.getMatch(2).getStartPos());
    assertEquals(7, kr.getMatch(2).getEndPos());
    assertEquals(5, kr.getMatch(3).getStartPos());
    assertEquals(7, kr.getMatch(3).getEndPos());
    // match nps as relation targets (right side)
    // return all nps whose children are articles
    SpanSegmentQuery ssq2 = new SpanSegmentQuery(sfq, scq2);
    kr = ki.search(ssq2, (short) 10);
    assertEquals((long) 2, kr.getTotalResults());
    assertEquals(2, kr.getMatch(0).getStartPos());
    assertEquals(4, kr.getMatch(0).getEndPos());
    assertEquals(5, kr.getMatch(1).getStartPos());
    assertEquals(7, kr.getMatch(1).getEndPos());
    // NOTE: cannot match the span id of the embedded relation
    // target and the np. So the segment match here is only by
    // positions. All nps in the same position will be matches.
    // return the articles whos parents are nps
    SpanFocusQuery sfq2 = new SpanFocusQuery(ssq2, (byte) 1);
    sfq2.setSorted(false);
    assertEquals("focus(1: spanSegment(focus(2: spanSegment({1: source:" + "{2: target:spanRelation(base:>:child-of)}}, " + "{1: spanTermWithId(base:pos:ART)})), " + "{2: <base:np />}))", sfq2.toString());
    kr = ki.search(sfq2, (short) 10);
    assertEquals((long) 2, kr.getTotalResults());
    assertEquals(2, kr.getMatch(0).getStartPos());
    assertEquals(3, kr.getMatch(0).getEndPos());
    assertEquals(5, kr.getMatch(1).getStartPos());
    assertEquals(6, kr.getMatch(1).getEndPos());
// for (Match m : kr.getMatches()) {
// System.out.println(m.getStartPos() + " " + m.getEndPos());
// }
}
Also used : SpanTermWithIdQuery(de.ids_mannheim.korap.query.SpanTermWithIdQuery) SpanClassQuery(de.ids_mannheim.korap.query.SpanClassQuery) SpanTermQuery(org.apache.lucene.search.spans.SpanTermQuery) SpanRelationQuery(de.ids_mannheim.korap.query.SpanRelationQuery) SpanSegmentQuery(de.ids_mannheim.korap.query.SpanSegmentQuery) Term(org.apache.lucene.index.Term) SpanFocusQuery(de.ids_mannheim.korap.query.SpanFocusQuery) SpanElementQuery(de.ids_mannheim.korap.query.SpanElementQuery) Test(org.junit.Test)

Aggregations

SpanRelationQuery (de.ids_mannheim.korap.query.SpanRelationQuery)20 Term (org.apache.lucene.index.Term)19 Test (org.junit.Test)19 SpanTermQuery (org.apache.lucene.search.spans.SpanTermQuery)18 SpanFocusQuery (de.ids_mannheim.korap.query.SpanFocusQuery)15 SpanRelationMatchQuery (de.ids_mannheim.korap.query.SpanRelationMatchQuery)9 SpanElementQuery (de.ids_mannheim.korap.query.SpanElementQuery)8 SpanClassQuery (de.ids_mannheim.korap.query.SpanClassQuery)7 SpanAttributeQuery (de.ids_mannheim.korap.query.SpanAttributeQuery)5 SpanWithAttributeQuery (de.ids_mannheim.korap.query.SpanWithAttributeQuery)5 SpanQuery (org.apache.lucene.search.spans.SpanQuery)4 SpanTermWithIdQuery (de.ids_mannheim.korap.query.SpanTermWithIdQuery)3 KrillIndex (de.ids_mannheim.korap.KrillIndex)2 DistanceConstraint (de.ids_mannheim.korap.query.DistanceConstraint)2 SpanDistanceQuery (de.ids_mannheim.korap.query.SpanDistanceQuery)2 SpanNextQuery (de.ids_mannheim.korap.query.SpanNextQuery)2 SpanReferenceQuery (de.ids_mannheim.korap.query.SpanReferenceQuery)2 SpanSegmentQuery (de.ids_mannheim.korap.query.SpanSegmentQuery)2 SpanQueryWrapper (de.ids_mannheim.korap.query.wrap.SpanQueryWrapper)1 RegexQuery (org.apache.lucene.sandbox.queries.regex.RegexQuery)1