Merge pull request #1082 from geonetwork/DH-relevant-similar-datasets…

…-suggestions [Datahub]: Change related datasets suggestions to be more relevant
geonetwork · Feb 25, 2025 · 6d78023 · 6d78023
2 parents 97ab0d3 + 6d1db07
commit 6d78023
Show file tree

Hide file tree

Showing 5 changed files with 67 additions and 21 deletions.
diff --git a/apps/datahub-e2e/src/e2e/datasetDetailPage.cy.ts b/apps/datahub-e2e/src/e2e/datasetDetailPage.cy.ts
@@ -676,14 +676,25 @@ describe('dataset pages', () => {
       })
       describe('related records', () => {
         beforeEach(() => {
-          cy.visit('/dataset/6d0bfdf4-4e94-48c6-9740-3f9facfd453c')
+          cy.visit('/dataset/a3774ef6-809d-4dd1-984f-9254f49cbd0a')
         })
         it('should display the related records', () => {
           cy.get('#related-records')
             .find('datahub-record-related-records')
             .find('gn-ui-related-record-card')
             .should('have.length.gt', 0)
         })
+        it('should display a similar related record', () => {
+          cy.get('#related-records')
+            .find('datahub-record-related-records')
+            .find('gn-ui-related-record-card')
+            .first()
+            .find('h4')
+            .should(
+              'have.text',
+              ` Metadata for E2E testing purpose. (this title is too long and should be cut, this title is too long and should be cut, this title is too long and should be cut, this title is too long and should be cut, this title is too long and should be cut) `
+            )
+        })
         it('goes to dataset on click', () => {
           let targetLink
           cy.get('#related-records')

diff --git a/libs/api/repository/src/lib/gn4/elasticsearch/elasticsearch.service.spec.ts b/libs/api/repository/src/lib/gn4/elasticsearch/elasticsearch.service.spec.ts
@@ -1,5 +1,8 @@
 import { ElasticsearchService } from './elasticsearch.service'
-import { elasticAggsResponseFixture } from '@geonetwork-ui/common/fixtures'
+import {
+  datasetRecordsFixture,
+  elasticAggsResponseFixture,
+} from '@geonetwork-ui/common/fixtures'
 import { LangService } from '@geonetwork-ui/util/i18n'
 import { EsSearchParams } from '@geonetwork-ui/api/metadata-converter'
 import { TestBed } from '@angular/core/testing'
@@ -678,9 +681,10 @@ describe('ElasticsearchService', () => {
   })
 
   describe('#getRelatedRecordPayload', () => {
+    const record = datasetRecordsFixture()[0]
     let payload
     beforeEach(() => {
-      payload = service.getRelatedRecordPayload('record title', 'some-uuid', 4)
+      payload = service.getRelatedRecordPayload(record, 4)
     })
     it('returns ES payload', () => {
       expect(payload).toEqual({
@@ -716,9 +720,33 @@ describe('ElasticsearchService', () => {
                   fields: [
                     'resourceTitleObject.default',
                     'resourceAbstractObject.default',
-                    'tag.raw',
+                    'allKeywords',
+                  ],
+                  like: [
+                    {
+                      doc: {
+                        resourceTitleObject: {
+                          default:
+                            'A very interesting dataset (un jeu de données très intéressant)',
+                        },
+                        resourceAbstractObject: {
+                          default: `# Introduction
+This dataset has been established for testing purposes.
+
+## Details
+This is a section about details. Here is an HTML tag: <img src="http://google.com" />. And [a link](https://google.com).
+
+## Informations intéressantes
+Cette section contient des *caractères internationaux* (ainsi que des "caractères spéciaux"). 'çàü^@/~^&`,
+                        },
+                        allKeywords: [
+                          'international',
+                          'test',
+                          '_another_keyword_',
+                        ],
+                      },
+                    },
                   ],
-                  like: 'record title',
                   max_query_terms: 12,
                   min_term_freq: 1,
                 },
@@ -734,7 +762,7 @@ describe('ElasticsearchService', () => {
                 },
               },
             ],
-            must_not: [{ wildcard: { uuid: 'some-uuid' } }],
+            must_not: [{ wildcard: { uuid: 'my-dataset-001' } }],
           },
         },
         size: 4,

diff --git a/libs/api/repository/src/lib/gn4/elasticsearch/elasticsearch.service.ts b/libs/api/repository/src/lib/gn4/elasticsearch/elasticsearch.service.ts
@@ -28,6 +28,7 @@ import {
 } from '@geonetwork-ui/api/metadata-converter'
 import { LangService } from '@geonetwork-ui/util/i18n'
 import { formatDate, isDateRange } from './date-range.utils'
+import { CatalogRecord } from '@geonetwork-ui/common/domain/model/record'
 
 export type DateRange = { start?: Date; end?: Date }
 
@@ -134,8 +135,7 @@ export class ElasticsearchService {
   }
 
   getRelatedRecordPayload(
-    title: string,
-    uuid: string,
+    record: CatalogRecord,
     size = 6,
     _source = [...ES_SOURCE_SUMMARY, 'allKeywords', 'createDate']
   ): EsSearchParams {
@@ -148,9 +148,23 @@ export class ElasticsearchService {
                 fields: [
                   'resourceTitleObject.default',
                   'resourceAbstractObject.default',
-                  'tag.raw',
+                  'allKeywords',
+                ],
+                like: [
+                  {
+                    doc: {
+                      resourceTitleObject: {
+                        default: record.title,
+                      },
+                      resourceAbstractObject: {
+                        default: record.abstract,
+                      },
+                      allKeywords: record.keywords.map(
+                        (keyword) => keyword.label
+                      ),
+                    },
+                  },
                 ],
-                like: title,
                 min_term_freq: 1,
                 max_query_terms: 12,
               },
@@ -166,7 +180,7 @@ export class ElasticsearchService {
               },
             },
           ],
-          must_not: [{ wildcard: { uuid: uuid } }],
+          must_not: [{ wildcard: { uuid: record.uniqueIdentifier } }],
         },
       },
       size,

diff --git a/libs/api/repository/src/lib/gn4/gn4-repository.spec.ts b/libs/api/repository/src/lib/gn4/gn4-repository.spec.ts
@@ -263,11 +263,8 @@ describe('Gn4Repository', () => {
       )
     })
     it('uses a related record ES payload', () => {
-      expect(gn4Helper.getRelatedRecordPayload).toHaveBeenCalledWith(
-        'A very interesting dataset (un jeu de données très intéressant)',
-        'my-dataset-001',
-        3
-      )
+      const record = datasetRecordsFixture()[0]
+      expect(gn4Helper.getRelatedRecordPayload).toHaveBeenCalledWith(record, 3)
     })
     it('returns the given results as records', () => {
       expect(results).toStrictEqual(datasetRecordsFixture())

diff --git a/libs/api/repository/src/lib/gn4/gn4-repository.ts b/libs/api/repository/src/lib/gn4/gn4-repository.ts
@@ -143,11 +143,7 @@ export class Gn4Repository implements RecordsRepositoryInterface {
         'bucket',
         null,
         JSON.stringify(
-          this.gn4SearchHelper.getRelatedRecordPayload(
-            similarTo.title,
-            similarTo.uniqueIdentifier,
-            3
-          )
+          this.gn4SearchHelper.getRelatedRecordPayload(similarTo, 3)
         )
       )
       .pipe(