def add_edges(self, max_citations=50):
"""
For each syllabus, register citation pairs as edges.
Args:
max_citations (int): Discard documents with > N citations.
"""
# Aggregate the CNs.
texts = (
fn.array_agg(HLOM_Record.control_number)
.coerce(False)
.alias('texts')
)
# Select syllabi and cited CNs.
documents = (
HLOM_Citation
.select(HLOM_Citation.document, texts)
.join(HLOM_Record)
.having(fn.count(HLOM_Record.id) <= max_citations)
.distinct(HLOM_Citation.document)
.group_by(HLOM_Citation.document)
)
for row in query_bar(documents):
for cn1, cn2 in combinations(row.texts, 2):
# If the edge exists, +1 the weight.
if self.graph.has_edge(cn1, cn2):
self.graph[cn1][cn2]['weight'] += 1
# Otherwise, initialize the edge.
else: self.graph.add_edge(cn1, cn2, weight=1)
def add_edges(self, max_texts=20):
"""
For each syllabus, register citation pairs as edges.
Args:
max_texts (int): Ignore docs with > than N citations.
"""
text_ids = (fn.array_agg(Text.id).coerce(False).alias('text_ids'))
docs = (Citation.select(Citation.document, text_ids).join(Text).having(
fn.count(Text.id) <= max_texts).where(Text.display == True).where(
Text.valid == True).group_by(Citation.document))
for row in query_bar(docs):
for tid1, tid2 in combinations(row.text_ids, 2):
# If the edge exists, increment the weight.
if self.graph.has_edge(tid1, tid2):
self.graph[tid1][tid2]['weight'] += 1
# Otherwise, initialize the edge.
else:
self.graph.add_edge(tid1, tid2, weight=1)
def add_edges(self):
"""
For each syllabus, register citation pairs as edges.
"""
text_ids = (
fn.array_agg(Text.id)
.coerce(False)
.alias('text_ids')
)
docs = (
Citation
.select(Citation.document, text_ids)
.join(Text)
.where(Text.display==True)
.where(Text.valid==True)
.group_by(Citation.document)
)
for row in query_bar(docs):
for tid1, tid2 in combinations(row.text_ids, 2):
# If the edge exists, increment the weight.
if self.graph.has_edge(tid1, tid2):
self.graph[tid1][tid2]['weight'] += 1
# Otherwise, initialize the edge.
else:
self.graph.add_edge(tid1, tid2, weight=1)
