mirror of https://github.com/OpenSPG/KAG
disable entity linking in postprocess by default (#304)
This commit is contained in:
parent
4ad5bded26
commit
1e57016373
|
@ -35,7 +35,7 @@ class KAGPostProcessor(PostProcessorABC):
|
||||||
|
|
||||||
def __init__(
|
def __init__(
|
||||||
self,
|
self,
|
||||||
similarity_threshold: float = 0.9,
|
similarity_threshold: float = None,
|
||||||
external_graph: ExternalGraphLoaderABC = None,
|
external_graph: ExternalGraphLoaderABC = None,
|
||||||
):
|
):
|
||||||
"""
|
"""
|
||||||
|
@ -180,8 +180,9 @@ class KAGPostProcessor(PostProcessorABC):
|
||||||
origin_num_nodes = len(input.nodes)
|
origin_num_nodes = len(input.nodes)
|
||||||
origin_num_edges = len(input.edges)
|
origin_num_edges = len(input.edges)
|
||||||
new_graph = self.filter_invalid_data(input)
|
new_graph = self.filter_invalid_data(input)
|
||||||
self.similarity_based_link(new_graph)
|
if self.similarity_threshold is not None:
|
||||||
self.external_graph_based_link(new_graph)
|
self.similarity_based_link(new_graph)
|
||||||
|
self.external_graph_based_link(new_graph)
|
||||||
new_num_nodes = len(new_graph.nodes)
|
new_num_nodes = len(new_graph.nodes)
|
||||||
new_num_edges = len(new_graph.edges)
|
new_num_edges = len(new_graph.edges)
|
||||||
logger.debug(
|
logger.debug(
|
||||||
|
|
|
@ -47,7 +47,6 @@ kag_builder_pipeline:
|
||||||
type: dict_reader # kag.builder.component.reader.dict_reader.DictReader
|
type: dict_reader # kag.builder.component.reader.dict_reader.DictReader
|
||||||
post_processor:
|
post_processor:
|
||||||
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
||||||
similarity_threshold: 0.9
|
|
||||||
splitter:
|
splitter:
|
||||||
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
||||||
split_length: 100000
|
split_length: 100000
|
||||||
|
|
|
@ -73,7 +73,6 @@ kag_builder_pipeline:
|
||||||
type: dict_reader # kag.builder.component.reader.dict_reader.DictReader
|
type: dict_reader # kag.builder.component.reader.dict_reader.DictReader
|
||||||
post_processor:
|
post_processor:
|
||||||
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
||||||
similarity_threshold: 0.9
|
|
||||||
splitter:
|
splitter:
|
||||||
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
||||||
split_length: 100000
|
split_length: 100000
|
||||||
|
|
|
@ -73,7 +73,6 @@ kag_builder_pipeline:
|
||||||
type: dict_reader # kag.builder.component.reader.dict_reader.DictReader
|
type: dict_reader # kag.builder.component.reader.dict_reader.DictReader
|
||||||
post_processor:
|
post_processor:
|
||||||
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
||||||
similarity_threshold: 0.9
|
|
||||||
splitter:
|
splitter:
|
||||||
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
||||||
split_length: 100000
|
split_length: 100000
|
||||||
|
|
|
@ -49,7 +49,6 @@ kag_builder_pipeline:
|
||||||
type: txt_reader # kag.builder.component.reader.txt_reader.TXTReader
|
type: txt_reader # kag.builder.component.reader.txt_reader.TXTReader
|
||||||
post_processor:
|
post_processor:
|
||||||
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
||||||
similarity_threshold: 0.9
|
|
||||||
splitter:
|
splitter:
|
||||||
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
||||||
split_length: 300
|
split_length: 300
|
||||||
|
|
|
@ -47,7 +47,6 @@ kag_builder_pipeline:
|
||||||
type: txt_reader # kag.builder.component.reader.txt_reader.TXTReader
|
type: txt_reader # kag.builder.component.reader.txt_reader.TXTReader
|
||||||
post_processor:
|
post_processor:
|
||||||
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
||||||
similarity_threshold: 0.9
|
|
||||||
splitter:
|
splitter:
|
||||||
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
||||||
split_length: 4950
|
split_length: 4950
|
||||||
|
|
|
@ -68,7 +68,6 @@ kag_builder_pipeline:
|
||||||
type: txt_reader # kag.builder.component.reader.text_reader.TXTReader
|
type: txt_reader # kag.builder.component.reader.text_reader.TXTReader
|
||||||
post_processor:
|
post_processor:
|
||||||
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
||||||
similarity_threshold: 0.9
|
|
||||||
external_graph: *external_graph_loader
|
external_graph: *external_graph_loader
|
||||||
splitter:
|
splitter:
|
||||||
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
||||||
|
|
|
@ -47,7 +47,6 @@ kag_builder_pipeline:
|
||||||
type: dict_reader # kag.builder.component.reader.dict_reader.DictReader
|
type: dict_reader # kag.builder.component.reader.dict_reader.DictReader
|
||||||
post_processor:
|
post_processor:
|
||||||
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
||||||
similarity_threshold: 0.9
|
|
||||||
splitter:
|
splitter:
|
||||||
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
||||||
split_length: 100000
|
split_length: 100000
|
||||||
|
|
|
@ -47,7 +47,6 @@ kag_builder_pipeline:
|
||||||
type: dict_reader # kag.builder.component.reader.dict_reader.DictReader
|
type: dict_reader # kag.builder.component.reader.dict_reader.DictReader
|
||||||
post_processor:
|
post_processor:
|
||||||
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
||||||
similarity_threshold: 0.9
|
|
||||||
splitter:
|
splitter:
|
||||||
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
||||||
split_length: 100000
|
split_length: 100000
|
||||||
|
|
|
@ -51,7 +51,6 @@ extract_runner:
|
||||||
name_col: title
|
name_col: title
|
||||||
post_processor:
|
post_processor:
|
||||||
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
||||||
similarity_threshold: 0.9
|
|
||||||
splitter:
|
splitter:
|
||||||
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
||||||
split_length: 100000
|
split_length: 100000
|
||||||
|
|
|
@ -47,7 +47,6 @@ kag_builder_pipeline:
|
||||||
type: dict_reader # kag.builder.component.reader.dict_reader.DictReader
|
type: dict_reader # kag.builder.component.reader.dict_reader.DictReader
|
||||||
post_processor:
|
post_processor:
|
||||||
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
type: kag_post_processor # kag.builder.component.postprocessor.kag_postprocessor.KAGPostProcessor
|
||||||
similarity_threshold: 0.9
|
|
||||||
splitter:
|
splitter:
|
||||||
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
type: length_splitter # kag.builder.component.splitter.length_splitter.LengthSplitter
|
||||||
split_length: 100000
|
split_length: 100000
|
||||||
|
|
Loading…
Reference in New Issue