graph-seeder 1.0.0.dev0__tar.gz → 1.0.0.dev1__tar.gz

{graph_seeder-1.0.0.dev0 → graph_seeder-1.0.0.dev1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: graph-seeder
-Version: 1.0.0.dev0
+Version: 1.0.0.dev1
 Summary: A powerful tool to extract and densify subgraphs from Knowledge Graphs via SPARQL or LMDB, with different extraction strategies.
 Requires-Python: >=3.9
 Requires-Dist: lmdb>=2.2.0

{graph_seeder-1.0.0.dev0 → graph_seeder-1.0.0.dev1}/pyproject.toml

@@ -7,7 +7,7 @@ packages = ["src/graph_seeder"]
 
 [project]
 name = "graph-seeder"
-version = "1.0.0.dev0"
+version = "1.0.0.dev1"
 description = "A powerful tool to extract and densify subgraphs from Knowledge Graphs via SPARQL or LMDB, with different extraction strategies."
 readme = "README.md"
 requires-python = ">=3.9"

{graph_seeder-1.0.0.dev0 → graph_seeder-1.0.0.dev1}/src/graph_seeder/SubgraphExtractor.py

@@ -281,10 +281,13 @@ class SubgraphExtractor:
     def densify_graph(
         self, triplets: list[tuple[str, str, str]], seeds_found: set[str]
     ):
+        explored_nodes = self.extractor_strategy.explored_nodes
+
         graph_connector: GraphConnector = GraphConnector(
             self.wrapper,
             self.uri_manager,
             self.extractor_strategy.graph,
+            explored_nodes,
             self.ui,
             self.cfg,
         )

{graph_seeder-1.0.0.dev0 → graph_seeder-1.0.0.dev1}/src/graph_seeder/densification/GraphConnector.py

@@ -18,6 +18,7 @@ class GraphConnector:
         wrapper: NeighborhoodWrapper,
         uri_manager: URIManager,
         graph: nx.MultiGraph,
+        explored_nodes: set[str],
         ui: ConsoleUI,
         config: dict,
     ):
@@ -33,7 +34,7 @@ class GraphConnector:
             config.get("densification", {}).get("mode", "most_connected").lower()
         )
         self.bfs: BidirectionalBFS = ComponentFactory.create_strategy(
-            wrapper, uri_manager, self.bfs_config
+            wrapper, uri_manager, self.bfs_config, explored_nodes=explored_nodes
         )
         self.bfs.load_graph(graph)
 

{graph_seeder-1.0.0.dev0 → graph_seeder-1.0.0.dev1}/src/graph_seeder/extraction/BFS/BFS.py

@@ -20,12 +20,12 @@ class BidirectionalBFS(ExtractionStrategy):
         wrapper: NeighborhoodWrapper,
         uri_manager: URIManager,
         cfg: dict,
+        explored_nodes: set[str] = None,
     ) -> None:
-        super().__init__(wrapper, uri_manager, cfg)
+        super().__init__(wrapper, uri_manager, cfg, explored_nodes)
         self.uri_manager = uri_manager
         self.cfg = cfg
 
-        self.explored_nodes: set[str] = set()
         self._has_path: bool = False
 
         self._excluded_nodes = set(cfg["graph_filters"]["exclude_nodes"])

{graph_seeder-1.0.0.dev0 → graph_seeder-1.0.0.dev1}/src/graph_seeder/extraction/ExtractionStrategy.py

@@ -14,13 +14,18 @@ logger = logging.getLogger("subgraph")
 
 class ExtractionStrategy(ABC):
     def __init__(
-        self, wrapper: NeighborhoodWrapper, uri_manager: URIManager, config: dict
+        self,
+        wrapper: NeighborhoodWrapper,
+        uri_manager: URIManager,
+        config: dict,
+        explored_nodes: set[str] = None,
     ):
         """Base class for extraction strategies that define how to extract a subgraph given a set of seed nodes."""
         self.wrapper = wrapper
         self.uri_manager = uri_manager
         self.config = config
         self.graph: MultiGraph = MultiGraph()
+        self.explored_nodes = set() if explored_nodes is None else explored_nodes
 
     @abstractmethod
     def extract(self, nodes: list[str]) -> tuple[list[tuple[str, str, str]], str]:

{graph_seeder-1.0.0.dev0 → graph_seeder-1.0.0.dev1}/src/graph_seeder/extraction/Hop/HopExpansion.py

@@ -11,9 +11,13 @@ class HopExpansion(ExtractionStrategy):
     """Simple expansion strategy that expands each node level by level up to max_hops."""
 
     def __init__(
-        self, wrapper: NeighborhoodWrapper, uri_manager: URIManager, config: dict
+        self,
+        wrapper: NeighborhoodWrapper,
+        uri_manager: URIManager,
+        config: dict,
+        explored_nodes: set[str] = None,
     ):
-        super().__init__(wrapper, uri_manager, config)
+        super().__init__(wrapper, uri_manager, config, explored_nodes)
 
         self.max_hops = config["extraction"]["max_hops"]
         self.excluded_nodes = set(
@@ -65,18 +69,37 @@ class HopExpansion(ExtractionStrategy):
             )
             next_level_nodes: set[str] = set()
 
-            for triplets in self.wrapper.get_neighborhood(list(current_level_nodes)):
-                for subj, pred, obj in triplets:
-                    self.graph.add_edge(subj, obj, key=pred)
-                    if subj in self.excluded_nodes or obj in self.excluded_nodes:
-                        continue
-
-                    all_triplets.add((subj, pred, obj))
-
-                    if subj not in visited_nodes:
-                        next_level_nodes.add(subj)
-                    if obj not in visited_nodes:
-                        next_level_nodes.add(obj)
+            nodes_to_query = []
+
+            for node in current_level_nodes:
+                if node in self.explored_nodes:
+                    if node in self.graph:
+                        for u, v, data in self.graph.edges(node, data=True):
+                            pred = data.get("key")
+                            neighbor = v if u == node else u
+
+                            if neighbor not in self.excluded_nodes:
+                                all_triplets.add((u, pred, v))
+                                if neighbor not in visited_nodes:
+                                    next_level_nodes.add(neighbor)
+                else:
+                    nodes_to_query.append(node)
+
+            if nodes_to_query:
+                for triplets in self.wrapper.get_neighborhood(nodes_to_query):
+                    for subj, pred, obj in triplets:
+                        self.graph.add_edge(subj, obj, key=pred)
+                        if subj in self.excluded_nodes or obj in self.excluded_nodes:
+                            continue
+
+                        all_triplets.add((subj, pred, obj))
+
+                        if subj not in visited_nodes:
+                            next_level_nodes.add(subj)
+                        if obj not in visited_nodes:
+                            next_level_nodes.add(obj)
+
+                self.explored_nodes.update(nodes_to_query)
 
             if not next_level_nodes:
                 logger.info("No more nodes to expand. Graph is fully explored.")

{graph_seeder-1.0.0.dev0 → graph_seeder-1.0.0.dev1}/src/graph_seeder/utils/Factory.py

@@ -1,3 +1,4 @@
+from graph_seeder.utils.URIManager import URIManager
 from graph_seeder.wrapper.NeighborhoodWrapper import NeighborhoodWrapper
 from graph_seeder.wrapper.sparql.GraphWrapper import GraphWrapper
 from graph_seeder.wrapper.sparql.client.SparqlClient import SparqlClient
@@ -47,7 +48,12 @@ class ComponentFactory:
         return wrapper_class(uri_manager, config)
 
     @staticmethod
-    def create_strategy(wrapper, uri_manager, config: dict):
+    def create_strategy(
+        wrapper: NeighborhoodWrapper,
+        uri_manager: URIManager,
+        config: dict,
+        explored_nodes=None,
+    ):
         """Create an extraction strategy based on the configuration."""
         strategy_type = config.get("extraction", {}).get("strategy", "bfs").lower()
 
@@ -61,4 +67,9 @@ class ComponentFactory:
                 f"Unknown strategy type: '{strategy_type}'. Valid options: {list(registered_strategies.keys())}"
             )
 
-        return registered_strategies[strategy_type](wrapper, uri_manager, config)
+        return registered_strategies[strategy_type](
+            wrapper,
+            uri_manager,
+            config,
+            explored_nodes,
+        )