C.J. Sullivan cj2001

## svo.json
{"type":"node","id":"0","labels":["Node"],"properties":{"node_labels":["Subject"],"name":"oh bah mə","word_vec":[-0.1991655146019209,-0.05351358562798536,0.9228555063544845,-0.6886134662870478,-0.48122510045452427,-0.40170435813501726,-0.8155066138337055,0.8357354634523664,-0.2347006396854041,0.26097340188742835,-0.2814595678676137,-0.9936133169296157,-0.3826829229304849,-0.9808776175986211,-0.20284189377753714,0.5015809495104013,0.16734307094472212,-0.5867540626284398,-0.04434894351605689,-0.6723292743695024,0.9054839794937932,-0.6772277556586848,0.38326882164488585,-0.7193810327293506,0.007985495945301846,-0.6796721226933016,0.359004027656604,0.3961584634066617,0.9789458680359799,0.3723659161768318,0.6861445958760459,0.39929387851793496,-0.3740958123352931,0.07034310049505343,0.5782312986452309,0.27844665689970527,0.04389886470438653,0.38707344473628447,-0.35045714426886465,-0.9373177969255766,-0.12407924520086833,-0.5629778221312765,-0.9617121312902264,-0.587069845901774,0.7978564493099858,0.92225653009432

## docker-compose.yml
version: '3.7'

services:
  neo4j:
    image: neo4j:4.2.3-enterprise
    container_name: "neo-gds1.5"
    volumes:
      - $HOME/graph_data/my_data:/data
      - $HOME/graph_data/my_data:/var/lib/neo4j/import
    ports:

## Dockerfile
FROM jupyter/datascience-notebook

COPY requirements.txt ./
RUN pip install -U pip
RUN pip install --no-cache-dir -r requirements.txt

ENV JUPYTER_ENABLE_LAB=yes

COPY --chown=${NB_UID}:${NB_GID} . /home/jovyan/work
WORKDIR /home/jovyan/work

## requirements.txt
jupyterlab==3.0.7
neo4j==4.2.1
py2neo==2021.0.1

## query_to_list_values.py
result = conn.query(query_string)
for record in result:
    print(record['c.category'], record['inDegree'])

## arxiv_query_db.py
query_string = '''
MATCH (c:Category)
RETURN c.category_name, SIZE(()-[:IN_CATEGORY]->(c)) AS inDegree
ORDER BY inDegree DESC LIMIT 20
'''
top_cat_df = pd.DataFrame([dict(_) for _ in conn.query(query_string)])
top_cat_df.head(20)

## add_category_and_author_nodes.py
categories = pd.DataFrame(df[['category_list']])
categories.rename(columns={'category_list':'category'},
                  inplace=True)
categories = categories.explode('category') \
                       .drop_duplicates(subset=['category'])

authors = pd.DataFrame(df[['cleaned_authors_list']])
authors.rename(columns={'cleaned_authors_list':'author'},
               inplace=True)
authors=authors.explode('author').drop_duplicates(subset=['author'])

## paper_nodes_and_edges.py
def add_papers(rows, batch_size=5000):
   # Adds paper nodes and (:Author)--(:Paper) and
   # (:Paper)--(:Category) relationships to the Neo4j graph as a
   # batch job.

   query = '''
   UNWIND $rows as row
   MERGE (p:Paper {id:row.id}) ON CREATE SET p.title = row.title

   // connect categories

## add_authors_categories.py
def add_categories(categories):
    # Adds category nodes to the Neo4j graph.
    query = '''
            UNWIND $rows AS row
            MERGE (c:Category {category: row.category})
            RETURN count(*) as total
            '''
    return conn.query(query, parameters = {'rows':categories.to_dict('records')})


## create_arxiv_constraints.py
conn.query('CREATE CONSTRAINT papers IF NOT EXISTS ON (p:Paper)     ASSERT p.id IS UNIQUE')
conn.query('CREATE CONSTRAINT authors IF NOT EXISTS ON (a:Author) ASSERT a.name IS UNIQUE')
conn.query('CREATE CONSTRAINT categories IF NOT EXISTS ON (c:Category) ASSERT c.category IS UNIQUE')
	version: '3.7'

	services:
	neo4j:
	image: neo4j:4.2.3-enterprise
	container_name: "neo-gds1.5"
	volumes:
	- $HOME/graph_data/my_data:/data
	- $HOME/graph_data/my_data:/var/lib/neo4j/import
	ports:
	FROM jupyter/datascience-notebook

	COPY requirements.txt ./
	RUN pip install -U pip
	RUN pip install --no-cache-dir -r requirements.txt

	ENV JUPYTER_ENABLE_LAB=yes

	COPY --chown=${NB_UID}:${NB_GID} . /home/jovyan/work
	WORKDIR /home/jovyan/work
	result = conn.query(query_string)
	for record in result:
	print(record['c.category'], record['inDegree'])
	query_string = '''
	MATCH (c:Category)
	RETURN c.category_name, SIZE(()-[:IN_CATEGORY]->(c)) AS inDegree
	ORDER BY inDegree DESC LIMIT 20
	'''
	top_cat_df = pd.DataFrame([dict(_) for _ in conn.query(query_string)])
	top_cat_df.head(20)
	categories = pd.DataFrame(df[['category_list']])
	categories.rename(columns={'category_list':'category'},
	inplace=True)
	categories = categories.explode('category') \
	.drop_duplicates(subset=['category'])

	authors = pd.DataFrame(df[['cleaned_authors_list']])
	authors.rename(columns={'cleaned_authors_list':'author'},
	inplace=True)
	authors=authors.explode('author').drop_duplicates(subset=['author'])
	def add_papers(rows, batch_size=5000):
	# Adds paper nodes and (:Author)--(:Paper) and
	# (:Paper)--(:Category) relationships to the Neo4j graph as a
	# batch job.

	query = '''
	UNWIND $rows as row
	MERGE (p:Paper {id:row.id}) ON CREATE SET p.title = row.title

	// connect categories
	def add_categories(categories):
	# Adds category nodes to the Neo4j graph.
	query = '''
	UNWIND $rows AS row
	MERGE (c:Category {category: row.category})
	RETURN count(*) as total
	'''
	return conn.query(query, parameters = {'rows':categories.to_dict('records')})
	conn.query('CREATE CONSTRAINT papers IF NOT EXISTS ON (p:Paper) ASSERT p.id IS UNIQUE')
	conn.query('CREATE CONSTRAINT authors IF NOT EXISTS ON (a:Author) ASSERT a.name IS UNIQUE')
	conn.query('CREATE CONSTRAINT categories IF NOT EXISTS ON (c:Category) ASSERT c.category IS UNIQUE')