| 1 |
17911
|
marek.hors
|
package eu.dnetlib.iis.export.actionmanager.module;
|
| 2 |
|
|
|
| 3 |
22191
|
marek.hors
|
import java.util.ArrayList;
|
| 4 |
|
|
import java.util.Collections;
|
| 5 |
21553
|
marek.hors
|
import java.util.List;
|
| 6 |
|
|
|
| 7 |
28145
|
marek.hors
|
import org.apache.hadoop.conf.Configuration;
|
| 8 |
28149
|
marek.hors
|
import org.apache.log4j.Logger;
|
| 9 |
28145
|
marek.hors
|
|
| 10 |
25095
|
marek.hors
|
import eu.dnetlib.actionmanager.actions.AtomicAction;
|
| 11 |
17911
|
marek.hors
|
import eu.dnetlib.actionmanager.common.Agent;
|
| 12 |
28141
|
marek.hors
|
import eu.dnetlib.data.mapreduce.util.OafDecoder;
|
| 13 |
22191
|
marek.hors
|
import eu.dnetlib.data.proto.KindProtos.Kind;
|
| 14 |
|
|
import eu.dnetlib.data.proto.OafProtos.Oaf;
|
| 15 |
|
|
import eu.dnetlib.data.proto.OafProtos.OafRel;
|
| 16 |
|
|
import eu.dnetlib.data.proto.RelTypeProtos.RelType;
|
| 17 |
28141
|
marek.hors
|
import eu.dnetlib.data.proto.RelTypeProtos.SubRelType;
|
| 18 |
|
|
import eu.dnetlib.data.proto.ResultResultProtos.ResultResult;
|
| 19 |
|
|
import eu.dnetlib.data.proto.ResultResultProtos.ResultResult.Similarity;
|
| 20 |
|
|
import eu.dnetlib.data.proto.ResultResultProtos.ResultResult.Similarity.Type;
|
| 21 |
28147
|
marek.hors
|
import eu.dnetlib.iis.common.WorkflowRuntimeParameters;
|
| 22 |
27589
|
marek.hors
|
import eu.dnetlib.iis.common.hbase.HBaseConstants;
|
| 23 |
21974
|
marek.hors
|
import eu.dnetlib.iis.documentssimilarity.schemas.DocumentSimilarity;
|
| 24 |
17911
|
marek.hors
|
|
| 25 |
|
|
/**
|
| 26 |
|
|
* {@link DocumentSimilarity} based action builder module.
|
| 27 |
|
|
* @author mhorst
|
| 28 |
|
|
*
|
| 29 |
|
|
*/
|
| 30 |
26236
|
marek.hors
|
public class DocumentSimilarityActionBuilderModuleFactory
|
| 31 |
|
|
implements ActionBuilderFactory<DocumentSimilarity> {
|
| 32 |
17911
|
marek.hors
|
|
| 33 |
28234
|
marek.hors
|
private static final AlgorithmName algorithmName = AlgorithmName.document_similarities_standard;
|
| 34 |
|
|
|
| 35 |
28149
|
marek.hors
|
private final Logger log = Logger.getLogger(this.getClass());
|
| 36 |
22684
|
marek.hors
|
|
| 37 |
26236
|
marek.hors
|
class DocumentSimilarityActionBuilderModule extends AbstractBuilderModule
|
| 38 |
|
|
implements ActionBuilderModule<DocumentSimilarity> {
|
| 39 |
|
|
|
| 40 |
28147
|
marek.hors
|
private final Float threshold;
|
| 41 |
|
|
|
| 42 |
26236
|
marek.hors
|
/**
|
| 43 |
|
|
* Default constructor.
|
| 44 |
|
|
* @param predefinedTrust
|
| 45 |
28147
|
marek.hors
|
* @param threshold similarity threshold, skipped when null
|
| 46 |
26236
|
marek.hors
|
*/
|
| 47 |
|
|
public DocumentSimilarityActionBuilderModule(
|
| 48 |
|
|
String predefinedTrust,
|
| 49 |
28147
|
marek.hors
|
Float threshold) {
|
| 50 |
30163
|
marek.hors
|
super(predefinedTrust, algorithmName);
|
| 51 |
28147
|
marek.hors
|
this.threshold = threshold;
|
| 52 |
22191
|
marek.hors
|
}
|
| 53 |
26236
|
marek.hors
|
|
| 54 |
|
|
@Override
|
| 55 |
30163
|
marek.hors
|
public List<AtomicAction> build(DocumentSimilarity object, Agent agent,
|
| 56 |
|
|
String actionSetId) {
|
| 57 |
26236
|
marek.hors
|
if (object==null) {
|
| 58 |
|
|
return Collections.emptyList();
|
| 59 |
|
|
}
|
| 60 |
28147
|
marek.hors
|
// checking similarity threshold if set
|
| 61 |
|
|
if (threshold!=null && object.getSimilarity()!=null &&
|
| 62 |
|
|
object.getSimilarity()<threshold) {
|
| 63 |
|
|
return Collections.emptyList();
|
| 64 |
|
|
}
|
| 65 |
|
|
// setting relations in both source and target objects
|
| 66 |
26236
|
marek.hors
|
List<AtomicAction> simActions = createActions(
|
| 67 |
|
|
object, actionSetId, agent, false);
|
| 68 |
|
|
List<AtomicAction> reverseSimActions = createActions(
|
| 69 |
|
|
object, actionSetId, agent, true);
|
| 70 |
|
|
List<AtomicAction> results = new ArrayList<AtomicAction>();
|
| 71 |
|
|
if (simActions!=null && !simActions.isEmpty()) {
|
| 72 |
|
|
results.addAll(simActions);
|
| 73 |
|
|
}
|
| 74 |
|
|
if (reverseSimActions!=null && !reverseSimActions.isEmpty()) {
|
| 75 |
|
|
results.addAll(reverseSimActions);
|
| 76 |
|
|
}
|
| 77 |
|
|
return results;
|
| 78 |
22191
|
marek.hors
|
}
|
| 79 |
26236
|
marek.hors
|
|
| 80 |
|
|
/**
|
| 81 |
|
|
* Creates similarity related puts.
|
| 82 |
|
|
* @param object
|
| 83 |
|
|
* @param actionSet
|
| 84 |
|
|
* @param agent
|
| 85 |
|
|
* @param backwardMode
|
| 86 |
|
|
* @return similarity related puts
|
| 87 |
|
|
*/
|
| 88 |
|
|
protected List<AtomicAction> createActions(DocumentSimilarity object,
|
| 89 |
|
|
String actionSet, Agent agent, boolean backwardMode) {
|
| 90 |
|
|
Oaf oafObjectRel = buildOAFRel(
|
| 91 |
|
|
object.getDocumentId().toString(),
|
| 92 |
|
|
object.getOtherDocumentId().toString(),
|
| 93 |
|
|
object.getSimilarity(), backwardMode);
|
| 94 |
|
|
if (oafObjectRel==null) {
|
| 95 |
|
|
return Collections.emptyList();
|
| 96 |
|
|
}
|
| 97 |
|
|
List<AtomicAction> actionList = new ArrayList<AtomicAction>();
|
| 98 |
|
|
AtomicAction currentAction = actionFactory.createAtomicAction(
|
| 99 |
|
|
actionSet, agent,
|
| 100 |
|
|
backwardMode?
|
| 101 |
|
|
object.getOtherDocumentId().toString():
|
| 102 |
|
|
object.getDocumentId().toString(),
|
| 103 |
31011
|
marek.hors
|
OafDecoder.decode(oafObjectRel).getCFQ(),
|
| 104 |
26236
|
marek.hors
|
backwardMode?
|
| 105 |
|
|
object.getDocumentId().toString():
|
| 106 |
|
|
object.getOtherDocumentId().toString(),
|
| 107 |
|
|
oafObjectRel.toByteArray());
|
| 108 |
|
|
actionList.add(currentAction);
|
| 109 |
|
|
return actionList;
|
| 110 |
22523
|
marek.hors
|
}
|
| 111 |
26236
|
marek.hors
|
|
| 112 |
|
|
/**
|
| 113 |
|
|
* Builds OAF object.
|
| 114 |
|
|
* @param source
|
| 115 |
|
|
* @param target
|
| 116 |
|
|
* @param score
|
| 117 |
|
|
* @param invert flag indicating source and target should be inverted
|
| 118 |
|
|
* @return OAF object
|
| 119 |
|
|
*/
|
| 120 |
|
|
private Oaf buildOAFRel(String sourceId, String targetDocId,
|
| 121 |
|
|
float score, boolean invert) {
|
| 122 |
|
|
OafRel.Builder relBuilder = OafRel.newBuilder();
|
| 123 |
|
|
if (!invert) {
|
| 124 |
|
|
relBuilder.setSource(sourceId);
|
| 125 |
|
|
relBuilder.setTarget(targetDocId);
|
| 126 |
|
|
|
| 127 |
|
|
} else {
|
| 128 |
|
|
relBuilder.setSource(targetDocId);
|
| 129 |
|
|
relBuilder.setTarget(sourceId);
|
| 130 |
|
|
}
|
| 131 |
28141
|
marek.hors
|
String relClass = invert?
|
| 132 |
|
|
Similarity.RelName.isAmongTopNSimilarDocuments.toString():
|
| 133 |
|
|
Similarity.RelName.hasAmongTopNSimilarDocuments.toString();
|
| 134 |
26236
|
marek.hors
|
relBuilder.setChild(false);
|
| 135 |
28141
|
marek.hors
|
relBuilder.setRelType(RelType.resultResult);
|
| 136 |
|
|
relBuilder.setSubRelType(SubRelType.similarity);
|
| 137 |
|
|
relBuilder.setRelClass(relClass);
|
| 138 |
|
|
ResultResult.Builder resultResultBuilder = ResultResult.newBuilder();
|
| 139 |
|
|
Similarity.Builder similarityBuilder = Similarity.newBuilder();
|
| 140 |
|
|
similarityBuilder.setRelMetadata(buildRelMetadata(
|
| 141 |
|
|
HBaseConstants.SEMANTIC_SCHEME_DNET_RELATIONS_RESULT_RESULT,
|
| 142 |
|
|
relClass));
|
| 143 |
|
|
similarityBuilder.setSimilarity(score);
|
| 144 |
|
|
similarityBuilder.setType(Type.STANDARD);
|
| 145 |
|
|
resultResultBuilder.setSimilarity(similarityBuilder.build());
|
| 146 |
|
|
relBuilder.setResultResult(resultResultBuilder.build());
|
| 147 |
|
|
|
| 148 |
|
|
Oaf.Builder oafBuilder = Oaf.newBuilder();
|
| 149 |
26236
|
marek.hors
|
oafBuilder.setKind(Kind.relation);
|
| 150 |
|
|
oafBuilder.setRel(relBuilder.build());
|
| 151 |
|
|
oafBuilder.setDataInfo(buildInference());
|
| 152 |
|
|
oafBuilder.setTimestamp(System.currentTimeMillis());
|
| 153 |
|
|
return oafBuilder.build();
|
| 154 |
22191
|
marek.hors
|
}
|
| 155 |
30163
|
marek.hors
|
|
| 156 |
|
|
@Override
|
| 157 |
|
|
public AlgorithmName getAlgorithName() {
|
| 158 |
|
|
return algorithmName;
|
| 159 |
|
|
}
|
| 160 |
22191
|
marek.hors
|
}
|
| 161 |
|
|
|
| 162 |
|
|
@Override
|
| 163 |
26236
|
marek.hors
|
public ActionBuilderModule<DocumentSimilarity> instantiate(
|
| 164 |
30163
|
marek.hors
|
String predefinedTrust, Configuration config) {
|
| 165 |
28147
|
marek.hors
|
String thresholdStr = config.get(
|
| 166 |
|
|
WorkflowRuntimeParameters.EXPORT_DOCUMENTSSIMILARITY_THRESHOLD);
|
| 167 |
|
|
Float threshold = null;
|
| 168 |
|
|
if (thresholdStr!=null && !WorkflowRuntimeParameters.UNDEFINED_NONEMPTY_VALUE.equals(
|
| 169 |
|
|
thresholdStr)) {
|
| 170 |
|
|
threshold = Float.valueOf(thresholdStr);
|
| 171 |
28149
|
marek.hors
|
log.warn("setting documents similarity exporter threshold to: " + threshold);
|
| 172 |
28147
|
marek.hors
|
}
|
| 173 |
26236
|
marek.hors
|
return new DocumentSimilarityActionBuilderModule(
|
| 174 |
30163
|
marek.hors
|
predefinedTrust, threshold);
|
| 175 |
22191
|
marek.hors
|
}
|
| 176 |
17911
|
marek.hors
|
}
|