Merge pull request #366 from weaviate/autocut

Add support for autocut

integration/test_graphql.py

@@ -20,6 +20,7 @@
                 {"dataType": ["string"], "description": "description", "name": "description"},
                 {"dataType": ["int"], "description": "size", "name": "size"},
             ],
+            "vectorizer": "text2vec-contextionary",
         }
     ]
 }
@@ -226,6 +227,19 @@ def test_hybrid_properties(client, properties: Optional[List[str]], num_results:
         assert len(result["data"]["Get"]["Ship"]) == 0
 
 
+@pytest.mark.parametrize("autocut,num_results", [(1, 1), (2, 6), (-1, len(SHIPS))])
+def test_autocut(client, autocut, num_results):
+    result = (
+        client.query.get("Ship", ["name"])
+        .with_hybrid(query="sponges", properties=["name", "description"], alpha=0.5)
+        .with_autocut(autocut)
+        .with_limit(len(SHIPS))
+        .do()
+    )
+    assert len(result["data"]["Get"]["Ship"]) == num_results
+    assert result["data"]["Get"]["Ship"][0]["name"] == "The Crusty Crab"
+
+
 def test_group_by(client, people_schema):
     """Test hybrid search with alpha=0.5 to have a combination of BM25 and vector search."""
     client.schema.delete_all()

weaviate/gql/get.py

@@ -177,8 +177,18 @@ def __init__(self, class_name: str, properties: Optional[PROPERTIES], connection
         self._group_by: Optional[GroupBy] = None
         self._alias: Optional[str] = None
         self._tenant: Optional[str] = None
+        self._autocut: Optional[int] = None
         self._consistency_level: Optional[ConsistencyLevel] = None
 
+    def with_autocut(self, autocut: int):
+        """Cuts off irrelevant results based on "jumps" in scores."""
+        if not isinstance(autocut, int):
+            raise TypeError("autocut must be of type int")
+
+        self._autocut = autocut
+        self._contains_filter = True
+        return self
+
     def with_tenant(self, tenant: str):
         """Sets a tenant for the query."""
         if not isinstance(tenant, str):
@@ -1190,6 +1200,8 @@ def build(self, wrap_get: bool = True) -> str:
                 query += self._consistency_level
             if self._tenant is not None:
                 query += f'tenant: "{self._tenant}"'
+            if self._autocut is not None:
+                query += f"autocut: {self._autocut}"
 
             query += ")"