Project

General

Profile

« Previous | Next » 

Revision 33091

Added by Marek Horst over 9 years ago

#577 renaming origins to matched and ingested

View differences:

modules/icm-iis-collapsers/trunk/src/main/java/eu/dnetlib/iis/collapsers/origins/PMCCitationCollapser.java
16 16 
 */
17 17 
public class PMCCitationCollapser extends OriginCollapser<CitationEnvelope, Citation> {
18 18 

  		




  19
  
  
    
	private final String ORIGIN_PMC_INGESTION = "pmc_ingestion";


  		




  20
  
  
    
	private final String ORIGIN_CERMINE = "cermine";


  		




  
  19
  
    
	private final String ORIGIN_INGESTED = "ingested";


  		




  
  20
  
    
	private final String ORIGIN_MATCHED = "matched";


  		




  21
  21
  
    
	


  		




  22
  22
  
    
    @Override


  		




  23
  23
  
    
    protected List<Citation> collapseBetweenOrigins(Map<String, List<Citation>> objects) {


  		




  ......




  26
  26
  
    
        } else if (objects.size()==2) {


  		




  27
  27
  
    
//        	important note: identifying cermine matches by referenceText


  		




  28
  28
  
    
        	Map<String,Citation> cermineCitationsMap = new HashMap<String, Citation>();


  		




  29
  
  
    
        	for (Citation currentCermineCitation : objects.get(ORIGIN_CERMINE)) {


  		




  
  29
  
    
        	for (Citation currentCermineCitation : objects.get(ORIGIN_MATCHED)) {


  		




  30
  30
  
    
        		if (currentCermineCitation.getEntry()!=null && 


  		




  31
  31
  
    
        				currentCermineCitation.getEntry().getRawText()!=null) {


  		




  32
  32
  
    
        			cermineCitationsMap.put(


  		




  ......




  35
  35
  
    
        		}


  		




  36
  36
  
    
        	}


  		




  37
  37
  
    
        	List<Citation> results = new ArrayList<Citation>();


  		




  38
  
  
    
        	for (Citation currentPmcCitation : objects.get(ORIGIN_PMC_INGESTION)) {


  		




  
  38
  
    
        	for (Citation currentPmcCitation : objects.get(ORIGIN_INGESTED)) {


  		




  39
  39
  
    
        		if (currentPmcCitation.getEntry()!=null && 


  		




  40
  40
  
    
        				currentPmcCitation.getEntry().getRawText()!=null) {


  		




  41
  41
  
    
        			Citation merged = merge(currentPmcCitation, 


  		




  ......




  54
  54
  
    
        } else {


  		




  55
  55
  
    
//        	we need to be strict to conduct collapsing process properly


  		




  56
  56
  
    
        	throw new RuntimeException("only two origins are supported: " + 


  		




  57
  
  
    
        			ORIGIN_PMC_INGESTION + " and " + ORIGIN_CERMINE + 


  		




  
  57
  
    
        			ORIGIN_INGESTED + " and " + ORIGIN_MATCHED + 


  		




  58
  58
  
    
        			" got: " + objects.keySet());


  		




  59
  59
  
    
        }


  		




  60
  60
  
    
    }


  		












  

    
      modules/icm-iis-collapsers/trunk/src/main/resources/eu/dnetlib/iis/collapsers/multiple_input_collapser/job.properties
    
  





  9
  9
  
    
#schema_input_envelope=eu.dnetlib.iis.metadataextraction.schemas.ExtractedDocumentMetadataEnvelope


  		




  10
  10
  
    

  		




  11
  11
  
    
#citation


  		




  12
  
  
    
origin_1=pmc_ingestion


  		




  
  12
  
    
origin_1=ingested


  		




  13
  13
  
    
input_1=/user/marek.horst/mainworkflows/primary/main/working_dir/transformers_citations_from_ingestpmc/output


  		




  14
  
  
    
origin_2=cermine


  		




  
  14
  
    
origin_2=matched


  		




  15
  15
  
    
input_2=/user/marek.horst/mainworkflows/primary/main/working_dir/transformers_citations_from_matching/output


  		




  16
  16
  
    
blocking_field=sourceDocumentId


  		




  17
  17
  
    
schema_input=eu.dnetlib.iis.common.citations.schemas.Citation


  		












Also available in:  Unified diff