fix: address review feedback for gender bias analysis

 - run gender bias analysis independently of AI consent and system availability
 - return NEUTRAL for empty bias results while reserving undefined for pending analysis
 - restore the previous legal scoring rules for CRITICAL_AGG and TRANSPARENCY
 - revert unrelated IntelliJ run configuration and persistent MySQL volume changes
 - add NON_EMPTY serialization to biased and compliance issue DTOs
 - validate the analysis request jobId with @NotNull
 - add guarded Liquibase preconditions for the biased-issue unique constraint
 - rename the gender_bias_score database column to ai_score
 - document the new JobRepository queries and explain their separate loading strategy
 - replace ineffective HashSet usage with ArrayList for compliance issues
 - remove unused gender analysis service and editor code
 - move calculateCombinedAiScore and its test to ComplianceScoreCalculator
 - add JavaDoc for the combined AI score calculation
 - cover empty gender analysis results in status and button tests
 - rename the stale codingDisplay test description
 - remove the ineffective language-change test
 - refactor editor tests to use component inputs and template events instead of private access

Author: ge94zec

.run/DocApplyApp.run.xml

@@ -2,10 +2,10 @@
   <configuration default="false" name="DocApplyApp" type="SpringBootApplicationConfigurationType" factoryName="Spring Boot">
     <option name="ACTIVE_PROFILES" value="dev" />
     <option name="FRAME_DEACTIVATION_UPDATE_POLICY" value="UpdateClassesAndResources" />
-    <module name="doc-apply.main" />
+    <module name="de.tum.cit.aet.doc-apply.main" />
     <option name="SPRING_BOOT_MAIN_CLASS" value="de.tum.cit.aet.DocApplyApp" />
     <method v="2">
       <option name="Make" enabled="true" />
     </method>
   </configuration>
-</component>
+</component>

docker/local-setup/mysql.yml

@@ -5,7 +5,6 @@ services:
     image: mysql:9.3.0
     volumes:
       - ./config/mysql:/etc/mysql/conf.d
-      - docapply-mysql-data:/var/lib/mysql
     environment:
       - MYSQL_ALLOW_EMPTY_PASSWORD=yes
       - MYSQL_DATABASE=docapply
@@ -19,7 +18,3 @@ services:
       interval: 5s
       timeout: 10s
       retries: 10
-
-volumes:
-  docapply-mysql-data:
-    name: docapply-mysql-data

docker/local-setup/services.yml

@@ -9,7 +9,3 @@ services:
     extends:
       file: keycloak.yml
       service: keycloak
-
-volumes:
-  docapply-mysql-data:
-    name: docapply-mysql-data

src/main/java/de/tum/cit/aet/ai/dto/AnalyzeJobDescriptionRequestDTO.java

@@ -1,5 +1,6 @@
 package de.tum.cit.aet.ai.dto;
 
+import jakarta.validation.constraints.NotNull;
 import java.util.UUID;
 
-public record AnalyzeJobDescriptionRequestDTO(UUID jobId, String title, String jobDescriptionEN, String jobDescriptionDE) {}
+public record AnalyzeJobDescriptionRequestDTO(@NotNull UUID jobId, String title, String jobDescriptionEN, String jobDescriptionDE) {}

src/main/java/de/tum/cit/aet/ai/dto/BiasedIssueDTO.java

@@ -1,8 +1,10 @@
 package de.tum.cit.aet.ai.dto;
 
+import com.fasterxml.jackson.annotation.JsonInclude;
 import de.tum.cit.aet.ai.domain.BiasedIssue;
 import de.tum.cit.aet.core.constants.GenderCategory;
 
+@JsonInclude(JsonInclude.Include.NON_EMPTY)
 public record BiasedIssueDTO(String language, String word, GenderCategory type) {
     public static BiasedIssueDTO from(BiasedIssue issue) {
         return new BiasedIssueDTO(issue.getLanguage(), issue.getWord(), issue.getType());

src/main/java/de/tum/cit/aet/ai/dto/ComplianceIssueDTO.java

@@ -1,9 +1,11 @@
 package de.tum.cit.aet.ai.dto;
 
+import com.fasterxml.jackson.annotation.JsonInclude;
 import de.tum.cit.aet.ai.constants.ComplianceAction;
 import de.tum.cit.aet.ai.constants.ComplianceCategory;
 import de.tum.cit.aet.ai.domain.ComplianceIssue;
 
+@JsonInclude(JsonInclude.Include.NON_EMPTY)
 public record ComplianceIssueDTO(
     String id,
     ComplianceCategory category,

src/main/java/de/tum/cit/aet/ai/service/GenderBiasAnalysisService.java

@@ -4,9 +4,7 @@
 import de.tum.cit.aet.core.constants.GenderCategory;
 import de.tum.cit.aet.core.service.GenderBiasAnalyzer;
 import java.util.ArrayList;
-import java.util.HashSet;
 import java.util.List;
-import java.util.Set;
 import lombok.RequiredArgsConstructor;
 import org.springframework.stereotype.Service;
 
@@ -19,17 +17,6 @@ public class GenderBiasAnalysisService {
 
     private final GenderBiasAnalyzer analyzer;
 
-    /**
-     * Analyze the given text for gender bias.
-     *
-     * @param text     the text to analyze
-     * @param language the language code (e.g., "en" or "de")
-     * @return a response containing the analysis result and identified biased words
-     */
-    public Set<BiasedIssue> analyzeText(String text, String language) {
-        return new HashSet<>(analyzeOccurrences(text, language));
-    }
-
     /**
      * Analyze the given text while retaining repeated occurrences for score calculation.
      *

src/main/java/de/tum/cit/aet/ai/util/ComplianceScoreCalculator.java

@@ -1,9 +1,12 @@
 package de.tum.cit.aet.ai.util;
 
 import de.tum.cit.aet.ai.constants.ComplianceCategory;
+import de.tum.cit.aet.ai.domain.ComplianceIssue;
 import de.tum.cit.aet.core.constants.GenderCategory;
+import java.util.HashSet;
 import java.util.List;
 import java.util.Map;
+import java.util.Set;
 import java.util.function.Function;
 import java.util.stream.Collectors;
 
@@ -38,21 +41,36 @@ public static int calculateLegalScore(List<ComplianceCategory> categories) {
             .stream()
             .collect(Collectors.groupingBy(Function.identity(), Collectors.counting()));
 
-        if (
-            counts.getOrDefault(ComplianceCategory.CRITICAL_AGG, 0L) > 0 ||
-            counts.getOrDefault(ComplianceCategory.DSGVO_MINIMIZATION, 0L) > 0
-        ) {
+        if (counts.getOrDefault(ComplianceCategory.CRITICAL_AGG, 0L) > 0) {
             return 0;
         }
 
-        double totalCount =
-            (double) counts.getOrDefault(ComplianceCategory.TRANSPARENCY, 0L) +
-            (double) counts.getOrDefault(ComplianceCategory.PUBLIC_SECTOR, 0L);
+        double totalCount = (double) counts.getOrDefault(ComplianceCategory.TRANSPARENCY, 0L);
 
         double score = 100.0 * Math.pow(PENALTY_FACTOR, totalCount);
         return (int) Math.max(0, Math.round(score));
     }
 
+    /**
+     * Combines the gender inclusivity score with the legal compliance score using
+     * their geometric mean. Compliance issues that represent the same finding in
+     * multiple languages are counted only once based on their non-empty identifier.
+     *
+     * @param genderScore the gender inclusivity score from 0 to 100
+     * @param complianceIssues the detected compliance issues across all languages
+     * @return the combined AI score from 0 to 100
+     */
+    public static int calculateCombinedAiScore(int genderScore, List<ComplianceIssue> complianceIssues) {
+        Set<String> issueIds = new HashSet<>();
+        int legalScore = calculateLegalScore(
+            complianceIssues.stream()
+                .filter(issue -> issue.getId() == null || issue.getId().isBlank() || issueIds.add(issue.getId()))
+                .map(ComplianceIssue::getCategory)
+                .toList()
+        );
+        return (int) Math.round(Math.sqrt((double) genderScore * legalScore));
+    }
+
     /**
      * Calculates the combined gender bias score across two languages for consistency.
      *

src/main/java/de/tum/cit/aet/job/domain/Job.java

@@ -112,7 +112,7 @@ public class Job extends AbstractAuditingEntity {
     private Set<Application> applications;
 
     // Compliance fields for score calculation
-    @Column(name = "gender_bias_score")
+    @Column(name = "ai_score")
     private Integer aiScore;
 
     @ElementCollection

src/main/java/de/tum/cit/aet/job/repository/JobRepository.java

@@ -1,5 +1,7 @@
 package de.tum.cit.aet.job.repository;
 
+import de.tum.cit.aet.ai.domain.BiasedIssue;
+import de.tum.cit.aet.ai.domain.ComplianceIssue;
 import de.tum.cit.aet.core.repository.DocApplyJpaRepository;
 import de.tum.cit.aet.job.constants.Campus;
 import de.tum.cit.aet.job.constants.JobState;
@@ -358,16 +360,47 @@ ORDER BY CONCAT(p.firstName, ' ', p.lastName) ASC
     @Query("SELECT DISTINCT j.image.imageId FROM Job j WHERE j.image.imageId IN :imageIds")
     Set<UUID> findInUseImageIds(@Param("imageIds") List<UUID> imageIds);
 
+    /**
+     * Loads a job with its supervising professor, research group and image.
+     * The issue collections are intentionally not part of the entity graph and
+     * are fetched by their own queries instead,
+     * since joining both would produce a Cartesian product.
+     *
+     * @param jobId the job identifier
+     * @return the job, if it exists
+     */
     @EntityGraph(attributePaths = { "supervisingProfessor", "researchGroup", "image" })
     @Query("SELECT j FROM Job j WHERE j.jobId = :jobId")
     Optional<Job> findByIdWithDetails(@Param("jobId") UUID jobId);
 
+    /**
+     * Loads the compliance issues of a job in a dedicated query. Fetching them together
+     * with the biased issues would produce a Cartesian product and duplicate list entries.
+     *
+     * @param jobId the job identifier
+     * @return the persisted compliance issues
+     */
     @Query("SELECT issue FROM Job j JOIN j.complianceIssues issue WHERE j.jobId = :jobId")
-    List<de.tum.cit.aet.ai.domain.ComplianceIssue> findComplianceIssuesByJobId(@Param("jobId") UUID jobId);
+    List<ComplianceIssue> findComplianceIssuesByJobId(@Param("jobId") UUID jobId);
 
+    /**
+     * Loads biased issues separately from compliance issues to avoid a Cartesian
+     * product and retain the set semantics of the persisted collection.
+     *
+     * @param jobId the job identifier
+     * @return the persisted biased issues
+     */
     @Query("SELECT issue FROM Job j JOIN j.biasedIssues issue WHERE j.jobId = :jobId")
-    Set<de.tum.cit.aet.ai.domain.BiasedIssue> findBiasedIssuesByJobId(@Param("jobId") UUID jobId);
+    Set<BiasedIssue> findBiasedIssuesByJobId(@Param("jobId") UUID jobId);
 
+    /**
+     * Loads the job used for an analysis update deliberately without an entity graph:
+     * the update only touches the score and the issue collections, so eagerly loading
+     * the professor, research group and image would be wasted work.
+     *
+     * @param jobId the job identifier
+     * @return the job to update, if it exists
+     */
     @Query("SELECT j FROM Job j WHERE j.jobId = :jobId")
     Optional<Job> findByIdForAiUpdate(@Param("jobId") UUID jobId);
 }