update(kag) update kag_config.yaml for examples

kag/bin/commands/info.py

@@ -58,14 +58,28 @@ def handle_cls(args: argparse.Namespace):
         availables = cls_obj.list_available_with_detail()
         seg = " " * 20
 
+        deduped_availables = {}
+        for register_name, cls_info in availables.items():
+            cls = cls_info["class"]
+            if cls not in deduped_availables:
+                deduped_availables[cls] = [register_name]
+            else:
+                deduped_availables[cls].append(register_name)
+
         print(f"{bold}{red}{seg}Documentation of {args.cls}{seg}{reset}")
         import inspect
 
         print(inspect.getdoc(cls_obj))
         print(f"{bold}{red}{seg}Registered subclasses of {args.cls}{seg}{reset}")
+        visited = set()
         for register_name, cls_info in availables.items():
-            print(f"{bold}{blue}[{cls_info['class']}]{reset}")
-            print(f"{bold}{green}Register Name:{reset} {register_name}\n")
+            cls = cls_info["class"]
+            if cls in visited:
+                continue
+            visited.add(cls)
+            print(f"{bold}{blue}[{cls}]{reset}")
+            register_names = " / ".join([f'"{x}"' for x in deduped_availables[cls]])
+            print(f"{bold}{green}Register Name:{reset} {register_names}\n")
 
             # print(f"Class Name: {cls_info['class']}")
             print(f"{bold}{green}Documentation:{reset}\n{cls_info['doc']}\n")

kag/builder/component/extractor/schema_constraint_extractor.py

@@ -31,6 +31,7 @@
 
 
 @ExtractorABC.register("schema_constraint")
+@ExtractorABC.register("schema_constraint_extractor")
 class SchemaConstraintExtractor(ExtractorABC):
     """
     Perform knowledge extraction for enforcing schema constraints, including entities, events and their edges.

kag/builder/component/extractor/schema_free_extractor.py

@@ -31,6 +31,7 @@
 
 
 @ExtractorABC.register("schema_free")
+@ExtractorABC.register("schema_free_extractor")
 class SchemaFreeExtractor(ExtractorABC):
     """
     A class for extracting knowledge graph subgraphs from text using a large language model (LLM).

kag/builder/component/mapping/spg_type_mapping.py

@@ -28,6 +28,7 @@
 
 
 @MappingABC.register("spg")
+@MappingABC.register("spg_mapping")
 class SPGTypeMapping(MappingABC):
     """
     A class for mapping SPG(Semantic-enhanced Programmable Graph) types and handling their properties and strategies.

kag/builder/component/mapping/spo_mapping.py

@@ -19,6 +19,7 @@
 
 
 @MappingABC.register("spo")
+@MappingABC.register("spo_mapping")
 class SPOMapping(MappingABC):
     """
     A class that extends the MappingABC base class.

kag/builder/component/postprocessor/kag_postprocessor.py

@@ -25,6 +25,7 @@
 
 
 @PostProcessorABC.register("base", as_default=True)
+@PostProcessorABC.register("kag_post_processor", as_default=True)
 class KAGPostProcessor(PostProcessorABC):
     """
     A class that extends the PostProcessorABC base class.

kag/builder/component/reader/dict_reader.py

@@ -17,6 +17,7 @@
 
 
 @ReaderABC.register("dict")
+@ReaderABC.register("dict_reader")
 class DictReader(ReaderABC):
     """
     A class for reading dictionaries into Chunk objects.

kag/builder/component/reader/txt_reader.py

@@ -20,6 +20,7 @@
 
 
 @ReaderABC.register("txt")
+@ReaderABC.register("txt_reader")
 class TXTReader(ReaderABC):
     """
     A class for parsing text files or text content into Chunk objects.

kag/builder/component/scanner/csv_scanner.py

@@ -18,6 +18,7 @@
 
 
 @ScannerABC.register("csv")
+@ScannerABC.register("csv_scanner")
 class CSVScanner(ScannerABC):
     def __init__(
         self,

kag/builder/component/scanner/dataset_scanner.py

@@ -20,6 +20,7 @@
 
 
 @ScannerABC.register("hotpotqa")
+@ScannerABC.register("hotpotqa_dataset_scanner")
 class HotpotqaCorpusScanner(ScannerABC):
     """
     A class for reading HotpotQA dataset and converting it into a list of dictionaries, inheriting from `ScannerABC`.
@@ -66,6 +67,8 @@ def load_data(self, input: Input, **kwargs) -> List[Output]:
 
 @ScannerABC.register("musique")
 @ScannerABC.register("2wiki")
+@ScannerABC.register("musique_dataset_scanner")
+@ScannerABC.register("2wiki_dataset_scanner")
 class MusiqueCorpusScanner(ScannerABC):
     """
     A class for reading Musique/2Wiki dataset and converting it into a list of dictionaries, inheriting from `ScannerABC`.

kag/builder/component/scanner/directory_scanner.py

@@ -19,6 +19,7 @@
 
 
 @ScannerABC.register("dir")
+@ScannerABC.register("dir_file_scanner")
 class DirectoryScanner(ScannerABC):
     """
     A class for reading files from a directory based on a specified file pattern or suffix, inheriting from `ScannerABC`.

kag/builder/component/splitter/length_splitter.py

@@ -20,6 +20,7 @@
 
 
 @SplitterABC.register("length")
+@SplitterABC.register("length_splitter")
 class LengthSplitter(BaseTableSplitter):
     """
     A class for splitting text based on length.

kag/builder/component/vectorizer/batch_vectorizer.py

@@ -130,6 +130,7 @@ def batch_generate(self, node_batch, batch_size=32):
 
 
 @VectorizerABC.register("batch")
+@VectorizerABC.register("batch_vectorizer")
 class BatchVectorizer(VectorizerABC):
     """
     A class for generating embedding vectors for node attributes in a SubGraph in batches.

kag/builder/component/writer/kg_writer.py

@@ -29,6 +29,7 @@ class AlterOperationEnum(str, Enum):
 
 
 @SinkWriterABC.register("kg", as_default=True)
+@SinkWriterABC.register("kg_writer", as_default=True)
 class KGWriter(SinkWriterABC):
     """
     A class for writing SubGraphs to a Knowledge Graph (KG) storage.

kag/builder/default_chain.py

@@ -29,6 +29,7 @@
 
 
 @KAGBuilderChain.register("structured")
+@KAGBuilderChain.register("structured_builder_chain")
 class DefaultStructuredBuilderChain(KAGBuilderChain):
     """
     A class representing a default SPG builder chain, used to import structured data based on schema definitions.
@@ -84,6 +85,7 @@ def build(self, **kwargs):
 
 
 @KAGBuilderChain.register("unstructured")
+@KAGBuilderChain.register("unstructured_builder_chain")
 class DefaultUnstructuredBuilderChain(KAGBuilderChain):
     """
     A class representing a default unstructured builder chain, used to build a knowledge graph from unstructured text data such as txt and pdf files.
@@ -169,7 +171,7 @@ def run_extract(chunk):
             for inner_future in tqdm(
                 as_completed(futures),
                 total=len(futures),
-                desc="Chunk Extraction",
+                desc="KAG Extraction From Chunk",
                 position=1,
                 leave=False,
             ):

kag/examples/2wiki/builder/indexer.py

@@ -20,7 +20,7 @@
 def buildKB(file_path):
     from kag.common.conf import KAG_CONFIG
 
-    runner = BuilderChainRunner.from_config(KAG_CONFIG.all_config["runner"])
+    runner = BuilderChainRunner.from_config(KAG_CONFIG.all_config["kag_builder_pipeline"])
     runner.invoke(file_path)
 
     logger.info(f"\n\nbuildKB successfully for {file_path}\n\n")