Merge pull request #180 from weaviate/hybrid/add-support-for-query-property-weights

Add `Bm25QueryProperty` type for use in weighted `.bm25` and `.hybrid`

Author: tsmith023

src/collections/query/integration.test.ts

@@ -94,6 +94,38 @@ describe('Testing of the collection.query methods with a simple collection', ()
     expect(ret.objects[0].uuid).toEqual(id);
   });
 
+  it('should query with bm25 and weighted query properties', async () => {
+    const ret = await collection.query.bm25('test', {
+      queryProperties: [
+        {
+          name: 'testProp',
+          weight: 2,
+        },
+        'testProp2',
+      ],
+    });
+    expect(ret.objects.length).toEqual(1);
+    expect(ret.objects[0].properties.testProp).toEqual('test');
+    expect(ret.objects[0].properties.testProp2).toEqual('test2');
+    expect(ret.objects[0].uuid).toEqual(id);
+  });
+
+  it('should query with bm25 and weighted query properties with a non-generic collection', async () => {
+    const ret = await client.collections.get(collectionName).query.bm25('test', {
+      queryProperties: [
+        {
+          name: 'testProp',
+          weight: 2,
+        },
+        'testProp2',
+      ],
+    });
+    expect(ret.objects.length).toEqual(1);
+    expect(ret.objects[0].properties.testProp).toEqual('test');
+    expect(ret.objects[0].properties.testProp2).toEqual('test2');
+    expect(ret.objects[0].uuid).toEqual(id);
+  });
+
   it('should query with hybrid', async () => {
     const ret = await collection.query.hybrid('test', { limit: 1 });
     expect(ret.objects.length).toEqual(1);

src/collections/query/types.ts

@@ -76,10 +76,18 @@ export type SearchOptions<T> = {
   returnReferences?: QueryReference<T>[];
 };
 
+/** Which property of the collection to perform the keyword search on. */
+export type Bm25QueryProperty<T> = {
+  /** The property name to search on. */
+  name: PrimitiveKeys<T>;
+  /** The weight to provide to the keyword search for this property. */
+  weight: number;
+};
+
 /** Base options available in the `query.bm25` method */
 export type BaseBm25Options<T> = SearchOptions<T> & {
   /** Which properties of the collection to perform the keyword search on. */
-  queryProperties?: PrimitiveKeys<T>[];
+  queryProperties?: (PrimitiveKeys<T> | Bm25QueryProperty<T>)[];
 };
 
 /** Options available in the `query.bm25` method when specifying the `groupBy` parameter. */
@@ -98,7 +106,7 @@ export type BaseHybridOptions<T> = SearchOptions<T> & {
   /** The specific vector to search for or a specific vector subsearch. If not specified, the query is vectorized and used in the similarity search. */
   vector?: NearVectorInputType | HybridNearTextSubSearch | HybridNearVectorSubSearch;
   /** The properties to search in. If not specified, all properties are searched. */
-  queryProperties?: PrimitiveKeys<T>[];
+  queryProperties?: (PrimitiveKeys<T> | Bm25QueryProperty<T>)[];
   /** The type of fusion to apply. If not specified, the default fusion type specified by the server is used. */
   fusionType?: 'Ranked' | 'RelativeScore';
   /** Specify which vector(s) to search on if using named vectors. */

src/collections/serialize/index.ts

@@ -68,11 +68,12 @@ import {
   PrimitiveFilterValueType,
   PrimitiveListFilterValueType,
 } from '../filters/types.js';
-import { MultiTargetVectorJoin } from '../index.js';
+import { MultiTargetVectorJoin, PrimitiveKeys } from '../index.js';
 import {
   BaseHybridOptions,
   BaseNearOptions,
   Bm25Options,
+  Bm25QueryProperty,
   FetchObjectByIdOptions,
   FetchObjectsOptions,
   HybridNearTextSubSearch,
@@ -358,12 +359,24 @@ export class Serialize {
     };
   };
 
+  private static bm25QueryProperties = <T>(
+    properties?: (PrimitiveKeys<T> | Bm25QueryProperty<T>)[]
+  ): string[] | undefined => {
+    return properties?.map((property) => {
+      if (typeof property === 'string') {
+        return property;
+      } else {
+        return `${property.name}^${property.weight}`;
+      }
+    });
+  };
+
   public static bm25 = <T>(args: { query: string } & Bm25Options<T>): SearchBm25Args => {
     return {
       ...Serialize.common(args),
       bm25Search: BM25.fromPartial({
         query: args.query,
-        properties: args.queryProperties,
+        properties: this.bm25QueryProperties(args.queryProperties),
       }),
       autocut: args.autoLimit,
     };
@@ -448,7 +461,7 @@ export class Serialize {
       hybridSearch: Hybrid.fromPartial({
         query: args.query,
         alpha: args.alpha ? args.alpha : 0.5,
-        properties: args.queryProperties,
+        properties: this.bm25QueryProperties(args.queryProperties),
         vectorBytes: vectorBytes,
         fusionType: fusionType(args.fusionType),
         targetVectors,