Revision 49029
Added by Claudio Atzori over 6 years ago
DedupUtils.java | ||
---|---|---|
4 | 4 |
|
5 | 5 |
import eu.dnetlib.data.mapreduce.Algorithms; |
6 | 6 |
import eu.dnetlib.data.mapreduce.JobParams; |
7 |
import eu.dnetlib.data.proto.DedupProtos.Dedup.RelName; |
|
8 |
import eu.dnetlib.data.proto.KindProtos.Kind; |
|
9 |
import eu.dnetlib.data.proto.OafProtos.Oaf; |
|
10 |
import eu.dnetlib.data.transform.xml.AbstractDNetXsltFunctions; |
|
11 |
import org.apache.hadoop.hbase.io.ImmutableBytesWritable; |
|
12 |
import org.apache.hadoop.hbase.util.Bytes; |
|
13 |
|
|
14 | 7 |
import eu.dnetlib.data.proto.DedupProtos.Dedup; |
15 | 8 |
import eu.dnetlib.data.proto.DedupSimilarityProtos.DedupSimilarity; |
16 | 9 |
import eu.dnetlib.data.proto.FieldTypeProtos.Qualifier; |
10 |
import eu.dnetlib.data.proto.KindProtos.Kind; |
|
11 |
import eu.dnetlib.data.proto.OafProtos.Oaf; |
|
17 | 12 |
import eu.dnetlib.data.proto.OafProtos.OafRel; |
18 | 13 |
import eu.dnetlib.data.proto.OafProtos.OafRel.Builder; |
19 | 14 |
import eu.dnetlib.data.proto.OrganizationOrganizationProtos.OrganizationOrganization; |
20 |
import eu.dnetlib.data.proto.PersonPersonProtos.PersonPerson; |
|
21 | 15 |
import eu.dnetlib.data.proto.RelMetadataProtos.RelMetadata; |
22 | 16 |
import eu.dnetlib.data.proto.RelTypeProtos.RelType; |
23 | 17 |
import eu.dnetlib.data.proto.RelTypeProtos.SubRelType; |
24 | 18 |
import eu.dnetlib.data.proto.ResultResultProtos.ResultResult; |
25 | 19 |
import eu.dnetlib.data.proto.TypeProtos.Type; |
20 |
import eu.dnetlib.data.transform.xml.AbstractDNetXsltFunctions; |
|
26 | 21 |
import eu.dnetlib.pace.config.DedupConfig; |
22 |
import org.apache.hadoop.hbase.io.ImmutableBytesWritable; |
|
23 |
import org.apache.hadoop.hbase.util.Bytes; |
|
27 | 24 |
|
28 | 25 |
public class DedupUtils { |
29 | 26 |
|
... | ... | |
121 | 118 |
switch (type) { |
122 | 119 |
case organization: |
123 | 120 |
return RelType.organizationOrganization; |
124 |
case person: |
|
125 |
return RelType.personPerson; |
|
126 | 121 |
case result: |
127 | 122 |
return RelType.resultResult; |
128 | 123 |
default: |
... | ... | |
147 | 142 |
oafRel.setOrganizationOrganization(OrganizationOrganization.newBuilder().setDedup( |
148 | 143 |
dedup(relClass.name(), "dnet:organization_organization_relations"))); |
149 | 144 |
break; |
150 |
case person: |
|
151 |
oafRel.setPersonPerson(PersonPerson.newBuilder().setDedup(DedupUtils.dedup(relClass.name(), "dnet:person_person_relations"))); |
|
152 |
break; |
|
153 | 145 |
case result: |
154 | 146 |
oafRel.setResultResult(ResultResult.newBuilder().setDedup(DedupUtils.dedup(relClass.name(), "dnet:result_result_relations"))); |
155 | 147 |
break; |
... | ... | |
168 | 160 |
oafRel.setOrganizationOrganization(OrganizationOrganization.newBuilder().setDedupSimilarity( |
169 | 161 |
dedupSimilarity(isSimilarTo, "dnet:organization_organization_relations"))); |
170 | 162 |
break; |
171 |
case person: |
|
172 |
oafRel.setPersonPerson(PersonPerson.newBuilder().setDedupSimilarity(dedupSimilarity(isSimilarTo, "dnet:person_person_relations"))); |
|
173 |
break; |
|
174 | 163 |
case result: |
175 | 164 |
oafRel.setResultResult(ResultResult.newBuilder().setDedupSimilarity(dedupSimilarity(isSimilarTo, "dnet:result_result_relations"))); |
176 | 165 |
break; |
Also available in: Unified diff
getting rid of person entities