Merge pull request #1506 from projectblacklight/pf-boost-2

Boost relevance scoring when search terms appear in close proximity. Followup to #532.
projectblacklight · Dec 19, 2023 · 4dbe26b · 4dbe26b
2 parents 7922386 + 4e6d9a8
commit 4dbe26b
Show file tree

Hide file tree

Showing 3 changed files with 57 additions and 24 deletions.
diff --git a/solr/conf/solrconfig.xml b/solr/conf/solrconfig.xml
@@ -93,6 +93,11 @@
             http://wiki.apache.org/solr/LocalParams
        -->
 
+       <!-- In general, matches using the pf parameter should be boosted higher
+            than qf, so the proximity of multiple search terms in the matching
+            documents is significant in relevance scoring. See:
+            https://solr.apache.org/guide/solr/latest/query-guide/dismax-query-parser.html#pf-phrase-fields-parameter
+       -->
        <str name="qf">
          collection_title_tesim^150
          title_tesim^100
@@ -110,27 +115,27 @@
          text
        </str>
        <str name="pf">
-         collection_title_tesim^150
-         title_tesim^100
-         normalized_title_teim^100
-         unitid_identifier_match^40
-         name_teim^10
-         place_teim^10
-         subject_teim^2
-         id
-         ead_ssi
-         ref_ssm
-         unitid_ssm
-         container_teim
-         parent_unittitles_tesim
-         text
+         collection_title_tesim^300
+         title_tesim^200
+         normalized_title_teim^200
+         unitid_identifier_match^80
+         name_teim^20
+         place_teim^20
+         subject_teim^5
+         id^2
+         ead_ssi^2
+         ref_ssm^2
+         unitid_ssm^2
+         container_teim^2
+         parent_unittitles_tesim^2
+         text^2
        </str>
 
        <str name="qf_container">
          container_teim
        </str>
        <str name="pf_container">
-         container_teim
+         container_teim^2
        </str>
        <str name="qf_identifier">
          id
@@ -139,36 +144,36 @@
          unitid_ssm
        </str>
        <str name="pf_identifier">
-         id
-         ead_ssi
-         ref_ssm
-         unitid_ssm
+         id^2
+         ead_ssi^2
+         ref_ssm^2
+         unitid_ssm^2
        </str>
        <str name="qf_name">
          name_teim
        </str>
        <str name="pf_name">
-         name_teim
+         name_teim^2
        </str>
        <str name="qf_place">
          place_teim
        </str>
        <str name="pf_place">
-         place_teim
+         place_teim^2
        </str>
        <str name="qf_subject">
          subject_teim
        </str>
        <str name="pf_subject">
-         subject_teim
+         subject_teim^2
        </str>
        <str name="qf_title">
          title_tesim
          normalized_title_teim
        </str>
        <str name="pf_title">
-         title_tesim
-         normalized_title_teim
+         title_tesim^2
+         normalized_title_teim^2
        </str>
 
        <int name="ps">3</int>

diff --git a/spec/features/search_query_spec.rb b/spec/features/search_query_spec.rb
@@ -69,4 +69,22 @@
       end
     end
   end
+
+  context 'when two terms match two docs but proximity differs (pf test)' do
+    it 'counts the doc where the terms are in close proximity as more relevant' do
+      visit search_catalog_path q: 'splendiferous escapades', search_field: 'all_fields'
+      within('.document-position-1') do
+        expect(page).to have_css '.al-document-abstract-or-scope',
+                                 text: /This will test the splendiferous escapades phrase/
+      end
+    end
+
+    it 'counts the doc where the terms are are far apart as less relevant' do
+      visit search_catalog_path q: 'splendiferous escapades', search_field: 'all_fields'
+      within('.document-position-2') do
+        expect(page).to have_css '.al-document-abstract-or-scope',
+                                 text: /This splendiferous test will help/
+      end
+    end
+  end
 end
diff --git a/spec/fixtures/ead/nlm/alphaomegaalpha.xml b/spec/fixtures/ead/nlm/alphaomegaalpha.xml
@@ -680,6 +680,11 @@
             <container id="aspace_b60ff470b57285b4f198c8a7edcf2f61"
               parent="aspace_1579b968f44866f4066a9387e4446ce1" type="folder">15</container>
           </did>
+          <scopecontent id="aspace_7482a2ddff80393b0c12a926f9bd6ee1">
+            <head>Scope and Contents Phrase Query (pf) test 1</head>
+            <p>This will test the splendiferous escapades phrase to help ensure that relevance
+              is impacted by the proximity of search terms.</p>
+          </scopecontent>
         </c>
         <c id="aspace_d908ca557e8b2f6a4d13002616b086c1" level="otherlevel">
           <did>
@@ -688,6 +693,11 @@
             <container id="aspace_6d4dfa7db9a186576147009f3d5bd9ee" label="Mixed Materials"
               type="folder">1</container>
           </did>
+          <scopecontent id="aspace_98ae59df3f802993a01476bc133bae04">
+            <head>Scope and Contents Phrase Query (pf) test 2</head>
+            <p>This splendiferous test will help to ensure that the relevance
+              is impacted by the proximity of search term phrase escapades.</p>
+          </scopecontent>
         </c>
         <c id="aspace_a68ee517caa6d765f01eb923c177af2b" level="otherlevel">
           <did>