def setUp(self):
self.fact1 = Fact(
"corpus1",
"corpus_type",
"timestamp_from",
"timestamp_to",
"timestamp_type",
"analysis_type",
"result_key",
"result_value",
"outlierness",
)
self.message1 = Message(self.fact1, 0.1, 0.2, 0.3)
self.fact2 = Fact(
"corpus2",
"corpus_type",
"timestamp_from",
"timestamp_to",
"timestamp_type",
"analysis_type",
"result_key",
"result_value",
"outlierness",
)
self.message2 = Message(self.fact2, 0.1, 0.2, 0.3)
self.document_plan_node = DocumentPlanNode(
[self.message1, self.message2], Relation.ELABORATION)
def run(
self, registry: Registry, random: Generator, language: str,
scored_messages: List[Message]
) -> Tuple[DocumentPlanNode, List[Message]]:
log.debug("Creating body document plan")
# Root contains a sequence of children
document_plan = DocumentPlanNode(children=[],
relation=Relation.SEQUENCE)
available_messages = scored_messages[:] # Make a copy s.t. we can modify in place
selected_nuclei: List[Message] = []
while True:
nucleus: Message
nucleus_score: float
nucleus, nucleus_score = self.select_next_nucleus(
available_messages, selected_nuclei)
if (nucleus_score < self.new_paragraph_absolute_threshold
or nucleus_score <
self.new_paragraph_relative_threshold(selected_nuclei)):
if selected_nuclei:
return (document_plan, scored_messages)
selected_nuclei.append(nucleus)
# Messages are only allowed in the DP once
available_messages = [
m for m in available_messages if m != nucleus
]
# Get a suitable amount of satellites
satellites: List[Message] = self.select_satellites_for_nucleus(
nucleus, available_messages)
# Messages are only allowed in the DP once
available_messages = [
m for m in available_messages if m not in satellites
]
document_plan.children.append(
DocumentPlanNode([nucleus] + satellites, Relation.SEQUENCE))
def run(self, registry: Registry, random: Generator, language: str,
scored_messages) -> Tuple[DocumentPlanNode, List[Message]]:
"""
Run this pipeline component.
"""
log.debug("Creating headline document plan")
# Root contains a sequence of children
document_plan = DocumentPlanNode(children=[],
relation=Relation.SEQUENCE)
headline_message, _ = self.select_next_nucleus(scored_messages, [])
all_messages = scored_messages
document_plan.children.append(
DocumentPlanNode(children=[headline_message],
relation=Relation.SEQUENCE))
return document_plan, all_messages