|
| 1 | +"""Integration tests for hybrid search + MMR diversity selection. |
| 2 | +
|
| 3 | +``diversity_selection`` is a hybrid-level argument: the server applies it as a |
| 4 | +post-fusion MMR pass (Weaviate >= 1.38.6). These tests assert that ``balance=0`` |
| 5 | +(pure diversity) produces a different ordering than ``balance=1`` (pure |
| 6 | +relevance), and that ``mmr.limit`` caps the result count. |
| 7 | +
|
| 8 | +The equivalent ``near_vector`` behaviour is covered in |
| 9 | +``test_collection_diversity.py``. |
| 10 | +""" |
| 11 | + |
| 12 | +import pytest |
| 13 | + |
| 14 | +from integration.conftest import CollectionFactory |
| 15 | +from weaviate.classes.query import Diversity, HybridVector |
| 16 | +from weaviate.collections.classes.config import Configure, DataType, Property |
| 17 | +from weaviate.collections.classes.data import DataObject |
| 18 | + |
| 19 | +MIN_VERSION = (1, 38, 6) |
| 20 | + |
| 21 | + |
| 22 | +def _skip_if_unsupported(collection) -> None: |
| 23 | + if collection._connection._weaviate_version.is_lower_than(*MIN_VERSION): |
| 24 | + pytest.skip("Hybrid diversity selection requires Weaviate >= 1.38.6") |
| 25 | + |
| 26 | + |
| 27 | +def _create_clustered_collection(collection_factory: CollectionFactory): |
| 28 | + """Create a collection with 3 tight clusters (a, b, c) of vectors in 3D.""" |
| 29 | + collection = collection_factory( |
| 30 | + properties=[Property(name="text", data_type=DataType.TEXT)], |
| 31 | + vectorizer_config=Configure.Vectorizer.none(), |
| 32 | + ) |
| 33 | + _skip_if_unsupported(collection) |
| 34 | + collection.data.insert_many( |
| 35 | + [ |
| 36 | + DataObject(properties={"text": "a1"}, vector=[1.0, 0.0, 0.0]), |
| 37 | + DataObject(properties={"text": "a2"}, vector=[0.95, 0.05, 0.0]), |
| 38 | + DataObject(properties={"text": "a3"}, vector=[0.9, 0.1, 0.0]), |
| 39 | + DataObject(properties={"text": "b1"}, vector=[0.0, 1.0, 0.0]), |
| 40 | + DataObject(properties={"text": "b2"}, vector=[0.05, 0.95, 0.0]), |
| 41 | + DataObject(properties={"text": "c1"}, vector=[0.0, 0.0, 1.0]), |
| 42 | + ] |
| 43 | + ) |
| 44 | + return collection |
| 45 | + |
| 46 | + |
| 47 | +def _create_large_collection(collection_factory: CollectionFactory, n_items: int = 50): |
| 48 | + """Create a collection with enough items (>25) that a small mmr.limit is distinguishable from the server's default limit.""" |
| 49 | + collection = collection_factory( |
| 50 | + properties=[Property(name="text", data_type=DataType.TEXT)], |
| 51 | + vectorizer_config=Configure.Vectorizer.none(), |
| 52 | + ) |
| 53 | + _skip_if_unsupported(collection) |
| 54 | + collection.data.insert_many( |
| 55 | + [ |
| 56 | + DataObject(properties={"text": f"t{i}"}, vector=[1.0 - 0.001 * i, 0.0, 0.0]) |
| 57 | + for i in range(n_items) |
| 58 | + ] |
| 59 | + ) |
| 60 | + return collection |
| 61 | + |
| 62 | + |
| 63 | +def test_hybrid_near_vector_balance_0_differs_from_balance_1( |
| 64 | + collection_factory: CollectionFactory, |
| 65 | +) -> None: |
| 66 | + """Hybrid near-vector: balance=0 (diversity) must reorder vs balance=1 (relevance).""" |
| 67 | + collection = _create_clustered_collection(collection_factory) |
| 68 | + balance_0 = collection.query.hybrid( |
| 69 | + query=None, |
| 70 | + vector=HybridVector.near_vector(vector=[1.0, 0.0, 0.0]), |
| 71 | + diversity_selection=Diversity.mmr(limit=3, balance=0.0), |
| 72 | + limit=3, |
| 73 | + ).objects |
| 74 | + balance_1 = collection.query.hybrid( |
| 75 | + query=None, |
| 76 | + vector=HybridVector.near_vector(vector=[1.0, 0.0, 0.0]), |
| 77 | + diversity_selection=Diversity.mmr(limit=3, balance=1.0), |
| 78 | + limit=3, |
| 79 | + ).objects |
| 80 | + assert [o.uuid for o in balance_0] != [o.uuid for o in balance_1] |
| 81 | + |
| 82 | + |
| 83 | +def test_hybrid_near_vector_balance_1_matches_baseline( |
| 84 | + collection_factory: CollectionFactory, |
| 85 | +) -> None: |
| 86 | + """Hybrid near-vector with MMR balance=1 (pure relevance) matches the plain baseline.""" |
| 87 | + collection = _create_clustered_collection(collection_factory) |
| 88 | + baseline = collection.query.hybrid( |
| 89 | + query=None, |
| 90 | + vector=HybridVector.near_vector(vector=[1.0, 0.0, 0.0]), |
| 91 | + limit=3, |
| 92 | + ).objects |
| 93 | + mmr_balance_1 = collection.query.hybrid( |
| 94 | + query=None, |
| 95 | + vector=HybridVector.near_vector(vector=[1.0, 0.0, 0.0]), |
| 96 | + diversity_selection=Diversity.mmr(limit=3, balance=1.0), |
| 97 | + limit=3, |
| 98 | + ).objects |
| 99 | + assert [o.uuid for o in baseline] == [o.uuid for o in mmr_balance_1] |
| 100 | + |
| 101 | + |
| 102 | +def test_hybrid_alpha_1_balance_0_differs_from_balance_1( |
| 103 | + collection_factory: CollectionFactory, |
| 104 | +) -> None: |
| 105 | + """Hybrid with explicit alpha=1.0 (pure vector) applies MMR like near_vector.""" |
| 106 | + collection = _create_clustered_collection(collection_factory) |
| 107 | + balance_0 = collection.query.hybrid( |
| 108 | + query="irrelevant", |
| 109 | + alpha=1.0, |
| 110 | + vector=HybridVector.near_vector(vector=[1.0, 0.0, 0.0]), |
| 111 | + diversity_selection=Diversity.mmr(limit=3, balance=0.0), |
| 112 | + limit=3, |
| 113 | + ).objects |
| 114 | + balance_1 = collection.query.hybrid( |
| 115 | + query="irrelevant", |
| 116 | + alpha=1.0, |
| 117 | + vector=HybridVector.near_vector(vector=[1.0, 0.0, 0.0]), |
| 118 | + diversity_selection=Diversity.mmr(limit=3, balance=1.0), |
| 119 | + limit=3, |
| 120 | + ).objects |
| 121 | + assert [o.uuid for o in balance_0] != [o.uuid for o in balance_1] |
| 122 | + |
| 123 | + |
| 124 | +def test_hybrid_respects_mmr_limit( |
| 125 | + collection_factory: CollectionFactory, |
| 126 | +) -> None: |
| 127 | + """Hybrid respects mmr.limit as the result-count cap when no outer limit is set.""" |
| 128 | + mmr_limit = 5 |
| 129 | + collection = _create_large_collection(collection_factory, n_items=50) |
| 130 | + |
| 131 | + result = collection.query.hybrid( |
| 132 | + query=None, |
| 133 | + vector=HybridVector.near_vector(vector=[1.0, 0.0, 0.0]), |
| 134 | + diversity_selection=Diversity.mmr(limit=mmr_limit, balance=0.5), |
| 135 | + ).objects |
| 136 | + assert len(result) == mmr_limit |
0 commit comments