Merge pull request #358 from wcmc-its/relationship_strategy#341

Gender strategy and misc improvements
wcmc-its · Jun 12, 2019 · dd4c22b · dd4c22b
2 parents 418bdce + 24c72ea
commit dd4c22b
Show file tree

Hide file tree

Showing 24 changed files with 476,103 additions and 215 deletions.
diff --git a/pom.xml b/pom.xml
@@ -258,17 +258,17 @@
 	    <dependency>
 	    	<groupId>edu.cornell.weill.reciter</groupId>
 	    	<artifactId>reciter-identity-model</artifactId>
-	    	<version>2.0.4</version>
+	    	<version>2.0.5</version>
 	    </dependency>
 	    <dependency>
 	    	<groupId>edu.cornell.weill.reciter</groupId>
 	    	<artifactId>reciter-article-model</artifactId>
-	    	<version>2.0.9</version>
+	    	<version>2.0.10</version>
 	    </dependency>
 	    <dependency>
 	    	<groupId>edu.cornell.weill.reciter</groupId>
 	    	<artifactId>reciter-dynamodb-model</artifactId>
-	    	<version>2.0.2</version>
+	    	<version>2.0.3</version>
 	    </dependency>
 	    <dependency>
 	    	<groupId>com.github.bohnman</groupId>

diff --git a/src/main/java/reciter/Application.java b/src/main/java/reciter/Application.java
@@ -52,17 +52,20 @@
 import com.google.common.collect.Iterables;
 
 import lombok.extern.slf4j.Slf4j;
+import reciter.database.dyanmodb.files.GenderFileImport;
 import reciter.database.dyanmodb.files.IdentityFileImport;
 import reciter.database.dyanmodb.files.InstitutionAfidFileImport;
 import reciter.database.dyanmodb.files.MeshTermFileImport;
 import reciter.database.dyanmodb.files.ScienceMetrixDepartmentCategoryFileImport;
 import reciter.database.dyanmodb.files.ScienceMetrixFileImport;
+import reciter.database.dynamodb.model.Gender;
 import reciter.database.dynamodb.model.InstitutionAfid;
 import reciter.database.dynamodb.model.MeshTerm;
 import reciter.database.dynamodb.model.ScienceMetrix;
 import reciter.database.dynamodb.model.ScienceMetrixDepartmentCategory;
 import reciter.engine.EngineParameters;
 import reciter.security.APIKey;
+import reciter.service.GenderService;
 import reciter.service.ScienceMetrixDepartmentCategoryService;
 import reciter.service.ScienceMetrixService;
 import reciter.service.dynamo.DynamoDbInstitutionAfidService;
@@ -98,6 +101,9 @@ public class Application {
     @Autowired
     private DynamoDbInstitutionAfidService dynamoDbInstitutionAfidService;
 
+    @Autowired
+    private GenderService genderService;
+
     @Value("${use.scopus.articles}")
     private boolean useScopusArticles;
 
@@ -232,6 +238,9 @@ public void loadDynamoDbTablesAfterStartUp() {
 			IdentityFileImport identityFileImport = ApplicationContextHolder.getContext().getBean(IdentityFileImport.class);
 			identityFileImport.importIdentity();
 
+			GenderFileImport genderFileImport = ApplicationContextHolder.getContext().getBean(GenderFileImport.class);
+			genderFileImport.importGender();
+
 			if(useScopusArticles) {
 				InstitutionAfidFileImport institutionAfidFileImport = ApplicationContextHolder.getContext().getBean(InstitutionAfidFileImport.class);
 				institutionAfidFileImport.importInstitutionAfids();
@@ -263,6 +272,12 @@ public void populateStaticEngineParameters() {
             }
             EngineParameters.setMeshCountMap(meshCountMap);
         }
+        log.info("Loading GenderProbability to Engine Parameters");
+        List<Gender> genders = genderService.findAll();
+        if(genders != null && !genders.isEmpty()) {
+        	EngineParameters.setGenders(genders);
+        }
+
         if(useScopusArticles) {
 	        log.info("Loading ScopusInstitutionalAfids to Engine Parameters");
 	        List<InstitutionAfid> instAfids = dynamoDbInstitutionAfidService.findAll();
@@ -271,6 +286,8 @@ public void populateStaticEngineParameters() {
 		        	EngineParameters.setAfiliationNameToAfidMap(institutionAfids);
 	        }
         }
+
+        log.info("ReCiter is up and ready to use. Please make sure its other components such as Pubmed-Retrieval-Tool is also setup if you wish to do retrieval.");
 	}
 
 

diff --git a/src/main/java/reciter/algorithm/cluster/article/scorer/ReCiterArticleScorer.java b/src/main/java/reciter/algorithm/cluster/article/scorer/ReCiterArticleScorer.java
@@ -35,6 +35,8 @@
 import reciter.algorithm.evidence.targetauthor.department.strategy.DepartmentStringMatchStrategy;
 import reciter.algorithm.evidence.targetauthor.email.EmailStrategyContext;
 import reciter.algorithm.evidence.targetauthor.email.strategy.EmailStringMatchStrategy;
+import reciter.algorithm.evidence.targetauthor.gender.GenderStrategyContext;
+import reciter.algorithm.evidence.targetauthor.gender.strategy.GenderStrategy;
 import reciter.algorithm.evidence.targetauthor.grant.GrantStrategyContext;
 import reciter.algorithm.evidence.targetauthor.grant.strategy.GrantStrategy;
 import reciter.algorithm.evidence.targetauthor.journalcategory.JournalCategoryStrategyContext;
@@ -150,9 +152,14 @@ public class ReCiterArticleScorer extends AbstractArticleScorer {
 	private StrategyContext personTypeStrategyContext;
 
 	/**
-	 * Accpeted Rejected .
+	 * Accepted Rejected .
 	 */
 	private StrategyContext acceptedRejectedStrategyContext;
+
+	/**
+	 * Gender Strategy
+	 */
+	private GenderStrategyContext genderStrategyContext;
 
 	/**
 	 * Remove clusters based on cluster information.
@@ -186,6 +193,7 @@ public ReCiterArticleScorer(Map<Long, ReCiterCluster> clusters, Identity identit
 		this.journalCategoryStrategyContext = new JournalCategoryStrategyContext(new JournalCategoryStrategy());
 		this.knownRelationshipsStrategyContext = new KnownRelationshipStrategyContext(new KnownRelationshipStrategy());
 		this.affiliationStrategyContext = new AffiliationStrategyContext(new CommonAffiliationStrategy());
+		this.genderStrategyContext = new GenderStrategyContext(new GenderStrategy());
 
 		// Using the following strategy contexts in sequence to reassign individual articles
 		// to selected clusters.
@@ -238,6 +246,10 @@ public ReCiterArticleScorer(Map<Long, ReCiterCluster> clusters, Identity identit
 		if(strategyParameters.isAverageClustering()) {
 			this.strategyContexts.add(this.averageClusteringStrategyContext);
 		}
+
+		if(strategyParameters.isGender()) {
+			this.strategyContexts.add(this.genderStrategyContext);
+		}
 	}
 
 
@@ -293,6 +305,10 @@ public void runArticleScorer(Map<Long, ReCiterCluster> clusters, Identity identi
 				((ReCiterArticleStrategyContext) acceptedRejectedStrategyContext).executeStrategy(reCiterArticles);
 			}
 
+			if(strategyParameters.isGender()) {
+				((TargetAuthorStrategyContext) genderStrategyContext).executeStrategy(reCiterArticles, identity);
+			}
+
 			if (strategyParameters.isAverageClustering()) {
 				((ClusterStrategyContext) averageClusteringStrategyContext).executeStrategy(entry.getValue());
 			}

diff --git a/...iter/algorithm/evidence/cluster/averageclustering/strategy/AverageClusteringStrategy.java b/...iter/algorithm/evidence/cluster/averageclustering/strategy/AverageClusteringStrategy.java
@@ -36,8 +36,9 @@ public double executeStrategy(ReCiterCluster reCiterCluster) {
 				if(ReCiterArticleScorer.strategyParameters.isUseGoldStandardEvidence()) {
 					double totalArticleScoreWithoutClustering =  ((reCiterArticle.getAuthorNameEvidence() != null)?(reCiterArticle.getAuthorNameEvidence().getNameScoreTotal()):0) +
 							((reCiterArticle.getEmailEvidence() != null)?reCiterArticle.getEmailEvidence().getEmailMatchScore():0) +
+							((reCiterArticle.getGenderEvidence() != null && reCiterArticle.getGenderEvidence().getGenderScoreIdentityArticleDiscrepancy() != null)?reCiterArticle.getGenderEvidence().getGenderScoreIdentityArticleDiscrepancy():0) +
 							reCiterArticle.getGrantEvidenceTotalScore() +
-							reCiterArticle.getRelationshipEvidencesTotalScore() +
+							((reCiterArticle.getRelationshipEvidence() != null)?reCiterArticle.getRelationshipEvidence().getRelationshipEvidenceTotalScore():0) +
 							((reCiterArticle.getEducationYearEvidence() != null)?reCiterArticle.getEducationYearEvidence().getDiscrepancyDegreeYearBachelorScore():0) +
 							((reCiterArticle.getEducationYearEvidence() != null)?reCiterArticle.getEducationYearEvidence().getDiscrepancyDegreeYearDoctoralScore():0) +
 							reCiterArticle.getOrganizationalEvidencesTotalScore() +
@@ -53,8 +54,9 @@ public double executeStrategy(ReCiterCluster reCiterCluster) {
 				} else {
 					double totalArticleScoreWithoutClustering = ((reCiterArticle.getAuthorNameEvidence() != null)?(reCiterArticle.getAuthorNameEvidence().getNameScoreTotal()):0) + 
 							((reCiterArticle.getEmailEvidence() != null)?reCiterArticle.getEmailEvidence().getEmailMatchScore():0) + 
+							((reCiterArticle.getGenderEvidence() != null && reCiterArticle.getGenderEvidence().getGenderScoreIdentityArticleDiscrepancy() != null)?reCiterArticle.getGenderEvidence().getGenderScoreIdentityArticleDiscrepancy():0) +
 							reCiterArticle.getGrantEvidenceTotalScore() +
-							reCiterArticle.getRelationshipEvidencesTotalScore() +
+							((reCiterArticle.getRelationshipEvidence() != null)?reCiterArticle.getRelationshipEvidence().getRelationshipEvidenceTotalScore():0) +
 							((reCiterArticle.getEducationYearEvidence() != null)?reCiterArticle.getEducationYearEvidence().getDiscrepancyDegreeYearBachelorScore():0) +
 							((reCiterArticle.getEducationYearEvidence() != null)?reCiterArticle.getEducationYearEvidence().getDiscrepancyDegreeYearDoctoralScore():0) +
 							reCiterArticle.getOrganizationalEvidencesTotalScore() +