feat(autocomplete_boost_exact_matches): improved boosting for exact m…

…atches
pelias · Oct 15, 2018 · 20764b8 · 20764b8
1 parent a6fc61b
commit 20764b8
Show file tree

Hide file tree

Showing 12 changed files with 140 additions and 16 deletions.
diff --git a/query/autocomplete.js b/query/autocomplete.js
@@ -40,7 +40,8 @@ query.score( peliasQuery.view.admin('locality') );
 query.score( peliasQuery.view.admin('neighbourhood') );
 
 // scoring boost
-query.score( views.boost_exact_matches );
+query.score( views.boost_exact_matches(false) );
+query.score( views.boost_exact_matches(true) );
 query.score( peliasQuery.view.focus( views.ngrams_strict ) );
 query.score( peliasQuery.view.popularity( views.pop_subquery ) );
 query.score( peliasQuery.view.population( views.pop_subquery ) );

diff --git a/query/view/boost_exact_matches.js b/query/view/boost_exact_matches.js
@@ -14,26 +14,58 @@ var peliasQuery = require('pelias-query'),
   the view uses some of the values from the 'search_defaults.js' file to add an
   additional 'SHOULD' condition which scores exact matches slighly higher
   than partial matches.
+
+  the 'includePartialTokens' variable was introduced in order to allow the view
+  to be reused as an additional boost for tokens which are in fact complete,
+  despite us not knowing for sure whether they are complete or not.
+
+  an example is 'Stop 2', without partial tokens the boost will only apply to
+  documents matching 'stop', with an additional view we can further boost
+  documents matching 'stop 2'.
+
+  note: it is most likely insufficent to include a version of this view in your
+  query which has includePartialTokens=true without also having a copy with
+  includePartialTokens=false. One view will boost the tokens that are known to
+  be complete and the other will additionally boost tokens which may or may not be
+  complete, as per the example above.
+
+  note: a clause has been included in the code which disables the view for
+  includePartialTokens=true if it would generate the exact same view as for
+  includePartialTokens=false.
 **/
 
-module.exports = function( vs ){
+module.exports = function( includePartialTokens ){
+  return function( vs ){
+
+    // make a copy of the variables so we don't interfere with the values
+    // passed to other views.
+    var vsCopy = new peliasQuery.Vars( vs.export() );
+
+    // copy phrase:* values from search defaults
+    vsCopy.var('phrase:analyzer').set(searchDefaults['phrase:analyzer']);
+    vsCopy.var('phrase:field').set(searchDefaults['phrase:field']);
+
+    // get a copy of only the *complete* tokens produced from the input:name
+    var tokens = vs.var('input:name:tokens_complete').get();
 
-  // make a copy of the variables so we don't interfere with the values
-  // passed to other views.
-  var vsCopy = new peliasQuery.Vars( vs.export() );
+    if( includePartialTokens ){
+      // get a copy of *all* tokens produced from the input:name (including partial tokens)
+      var allTokens = vs.var('input:name:tokens').get();
 
-  // copy phrase:* values from search defaults
-  vsCopy.var('phrase:analyzer').set(searchDefaults['phrase:analyzer']);
-  vsCopy.var('phrase:field').set(searchDefaults['phrase:field']);
+      // a duplicate view would be generated, fail now, don't render this view.
+      // see file comments for more info
+      if( allTokens.join(' ') === tokens.join(' ') ){ return null; }
 
-  // get a copy of the *complete* tokens produced from the input:name
-  var tokens = vs.var('input:name:tokens_complete').get();
+      // use *all* the tokens for this view instead of only the complete tokens.
+      tokens = allTokens;
+    }
 
-  // no valid tokens to use, fail now, don't render this view.
-  if( !tokens || tokens.length < 1 ){ return null; }
+    // no valid tokens to use, fail now, don't render this view.
+    if( !tokens || tokens.length < 1 ){ return null; }
 
-  // set 'input:name' to be only the fully completed characters
-  vsCopy.var('input:name').set( tokens.join(' ') );
+    // set 'input:name' to be only the fully completed characters
+    vsCopy.var('input:name').set( tokens.join(' ') );
 
-  return peliasQuery.view.phrase( vsCopy );
+    return peliasQuery.view.phrase( vsCopy );
+  };
 };
diff --git a/test/unit/fixture/autocomplete_boundary_country.js b/test/unit/fixture/autocomplete_boundary_country.js
@@ -25,6 +25,16 @@ module.exports = {
         }
       }],
       'should':[{
+        'match': {
+          'phrase.default': {
+            'analyzer': 'peliasPhrase',
+            'boost': 1,
+            'slop': 3,
+            'query': 'test',
+            'type': 'phrase'
+          }
+        }
+      },{
         'function_score': {
           'query': {
             'match_all': {}

diff --git a/test/unit/fixture/autocomplete_linguistic_bbox_san_francisco.js b/test/unit/fixture/autocomplete_linguistic_bbox_san_francisco.js
@@ -18,6 +18,16 @@ module.exports = {
         }
       }],
       'should':[{
+        'match': {
+          'phrase.default': {
+            'analyzer': 'peliasPhrase',
+            'type': 'phrase',
+            'boost': 1,
+            'slop': 3,
+            'query': 'test'
+          }
+        }
+      },{
         'function_score': {
           'query': {
             'match_all': {}

diff --git a/test/unit/fixture/autocomplete_linguistic_final_token.js b/test/unit/fixture/autocomplete_linguistic_final_token.js
@@ -22,6 +22,16 @@ module.exports = {
             'type': 'phrase'
           }
         }
+      },{
+        'match': {
+          'phrase.default': {
+            'analyzer': 'peliasPhrase',
+            'boost': 1,
+            'slop': 3,
+            'query': 'one t',
+            'type': 'phrase'
+          }
+        }
       },{
         'function_score': {
           'query': {

diff --git a/test/unit/fixture/autocomplete_linguistic_focus.js b/test/unit/fixture/autocomplete_linguistic_focus.js
@@ -18,6 +18,16 @@ module.exports = {
         }
       }],
       'should': [{
+        'match': {
+          'phrase.default': {
+            'analyzer': 'peliasPhrase',
+            'boost': 1,
+            'slop': 3,
+            'query': 'test',
+            'type': 'phrase'
+          }
+        }
+      },{
         'function_score': {
           'query': {
             'match': {

diff --git a/test/unit/fixture/autocomplete_linguistic_focus_null_island.js b/test/unit/fixture/autocomplete_linguistic_focus_null_island.js
@@ -18,6 +18,16 @@ module.exports = {
         }
       }],
       'should': [{
+        'match': {
+          'phrase.default': {
+            'analyzer': 'peliasPhrase',
+            'boost': 1,
+            'slop': 3,
+            'query': 'test',
+            'type': 'phrase'
+          }
+        }
+      },{
         'function_score': {
           'query': {
             'match': {

diff --git a/test/unit/fixture/autocomplete_linguistic_multiple_tokens.js b/test/unit/fixture/autocomplete_linguistic_multiple_tokens.js
@@ -40,6 +40,17 @@ module.exports = {
             }
           }
         },
+        {
+          'match': {
+            'phrase.default': {
+              'analyzer' : 'peliasPhrase',
+              'type' : 'phrase',
+              'boost' : 1,
+              'slop' : 3,
+              'query' : 'one two three'
+            }
+          }
+        },
         {
         'function_score': {
           'query': {

diff --git a/test/unit/fixture/autocomplete_linguistic_only.js b/test/unit/fixture/autocomplete_linguistic_only.js
@@ -18,6 +18,16 @@ module.exports = {
         }
       }],
       'should':[{
+        'match': {
+          'phrase.default': {
+            'analyzer': 'peliasPhrase',
+            'type': 'phrase',
+            'boost': 1,
+            'slop': 3,
+            'query': 'test'
+          }
+        }
+      },{
         'function_score': {
           'query': {
             'match_all': {}

diff --git a/test/unit/fixture/autocomplete_with_layer_filtering.js b/test/unit/fixture/autocomplete_with_layer_filtering.js
@@ -18,6 +18,16 @@ module.exports = {
         }
       }],
       'should':[{
+        'match': {
+          'phrase.default': {
+            'analyzer': 'peliasPhrase',
+            'boost': 1,
+            'slop': 3,
+            'query': 'test',
+            'type': 'phrase'
+          }
+        }
+      },{
         'function_score': {
           'query': {
             'match_all': {}

diff --git a/test/unit/fixture/autocomplete_with_source_filtering.js b/test/unit/fixture/autocomplete_with_source_filtering.js
@@ -18,6 +18,16 @@ module.exports = {
         }
       }],
       'should':[{
+        'match': {
+          'phrase.default': {
+            'analyzer': 'peliasPhrase',
+            'boost': 1,
+            'slop': 3,
+            'query': 'test',
+            'type': 'phrase'
+          }
+        }
+      },{
         'function_score': {
           'query': {
             'match_all': {}

diff --git a/test/unit/query/autocomplete.js b/test/unit/query/autocomplete.js
@@ -206,7 +206,7 @@ module.exports.tests.query = function(test, common) {
     var expected = require('../fixture/autocomplete_linguistic_bbox_san_francisco');
 
     t.deepEqual(compiled.type, 'autocomplete', 'query type set');
-    t.deepEqual(compiled.body, expected, 'autocomplete_linguistic_focus_null_island');
+    t.deepEqual(compiled.body, expected, 'autocomplete_linguistic_bbox_san_francisco');
     t.end();
   });
 };