Giters

xilacxilac / KnowledgeGraph

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Knowledge Graph

KnowledgeGraph.py

Description:

  • Provides a simple framework to build, populate, and query our RDF Graph

Attributes

def __init__(self) -> None:
  # Stores the graph as an RDF
  self.rdf = Graph()
  self.uri = "http://example.org/"

  # Stores nodes
  self.rdf_nodes: dict[str, URIRef] = dict()

  for types in rdf_types:
    self.rdf_nodes[types] = URIRef(self.uri + types)
  for risk_factor in risk_factors:
    self.rdf_nodes[f"riskFactor/{risk_factor}"] = URIRef(self.uri + f"riskFactor/{risk_factor}")
    self.rdf.add((self.rdf_nodes[f"riskFactor/{risk_factor}"], RDF.type, self.rdf_nodes["riskFactor"]))

  # Stores predicates (edges)
  self.predicates: dict[str, URIRef] = dict()

  for predicate in predicates:
    self.predicates[predicate] = URIRef(self.uri + predicate)

  # Counter Variables
  self.SEC_sentence_count = dict()
  • self.rdf: An RDF Graph which contains a set of RDF triples
  • self.uri: The Uniform Resource Identifier, an example is set in this case
  • self.rdf_nodes: Stores frequently used RDF nodes, ensures uniqueness and re-usability
  • self.predicates: Stores frequently used RDF predicates, ensures uniqueness and re-usability
  • self.SEC_sentence_count: a unique counter for each company to ensure unique sentence names

Common Nodes and Predicates

  • Node Types
    • streetAddress, city, state, country, company, sector, industry, riskFactor, year, month, day, SECSentence, industry, sector, longBusinessSummary, fullTimeEmployees, auditRisk, boardRisk, compensationRisk, shareHolderRightsRisk, overallRisk, symbol
  • Predicates
    • RDF.type, hasState, belongsToCountry, hasCity, belongsToState, hasAddress, belongsToCity, isYear, isMonth, text, relatedToRiskFactor, belongsToCompany, hasRiskFactor, companiesWithRiskFactor, hasFilingYear, hasSECSentence
  • Risk Factors
    • General, Weather, Political, Economy, Energy, Business

Schemas

  • Company
    • Object Type: RDF.type
    • Ticker: symbol
    • Location of HQ: headquarteredIn
    • Long Business Summary: longBusinessSummary
    • Full-time Employees: fullTimeEmployees
    • Audit Risk (1-10, low-high): auditRisk
    • Board Risk (1-10, low-high): boardRisk
    • Compensation Risk (1-10, low-high): compensationRisk
    • Shareholder Rights Risk (1-10, low-high): shareHolderRightsRisk
    • Overall Risk (1-10, low-high): overallRisk
    • Industry: industryKey
    • Sector: sectorKey
    • SEC 10-K Filing Sentence [Optional]: hasSECSentence
    • Risk Factors [Optional]: hasRiskFactor
  • SECSentence
    • Object Type: RDF.type
    • SEC Sentence: text
    • Filing Year: hasFilingYear
    • Related Risk Factor: relatedToRiskFactor
    • Related Company: belongsToCompany
  • Location - Address
    • Object Type: RDF.type
    • City: belongsToCity
    • State: belongsToState
    • Country: belongsToCountry
    • Companies at Address: companiesHeadquartedIn
  • Location - City
    • Object Type: RDF.type
    • State: belongsToState
    • Country: belongsToCountry
    • Companies in City: companiesHeadquartedIn
    • Address in City: hasAddress
  • Location - State
    • Object Type: RDF.type
    • Country: belongsToCountry
    • Companies in State: companiesHeadquartedIn
    • City in State: hasCity
  • Location - Country
    • Object Type: RDF.type
    • Companies in Country: companiesHeadquartedIn
    • State in Country: hasState

Construction Functions

  • format_location(location: tuple[str, str, str, str]) -> tuple[str, str, str, str]
    • Cleans and reformats a location to a format which rdflib can support
  • add_edge(self, node1: URIRef, node2: URIRef, relation1: str, relation2: str = "") -> None:
    • Creates an edge between two URIRefs
  • add_attribute_literal(self, node1: URIRef, attribute: str, relation1: str) -> None:
    • Creates an attribute edge for a URIRef
  • add_attribute_resource(self, node1: URIRef, attribute: str, relation1: str, relation2: str = "", namespace: str = "") -> None:
    • Same as add_edge(), but you can pass in a custom namespace
  • add_location(self, address: str, city: str, state: str, country: str):
    • Adds location nodes and edges
  • add_date(self, year: str, month: str, day: str):
    • Adds date nodes and edges
  • add_company(self, ticker: str) -> str:
    • Adds a company node with attributes with information from yfinance (attribute details in Schemas)
  • add_risk_factors(self, ticker: str) -> str:
    • Adds sentences in a company's SEC 10-K fillings that correspond with a risk factor
  • quick_build(self) -> str:
    • Iterates through a preset list of tickers and builds the graph

Query Functions

  • query_by_sector(self, sector):
    • Finds companies in a sector and returns a list of tuples with the company name and ticker
  • query_by_industries(self, industry):
    • Finds companies in an industry and returns a list of tuples with the company name and ticker
  • query_by_ticker(self, ticker):
    • Finds a company by ticker and returns all of its edges and corresponding nodes
  • query_risk_factors(self, ticker):
    • Finds a company by ticker and returns their risk factors
  • query_SEC_sentences(self, ticker):
    • Finds a company by ticker and returns all of their risk-related SEC 10-K filing sentences
  • query_SEC_sentences_by_year(self, ticker, year):
    • Finds a company by ticker and returns all of their risk-related SEC 10-K filing sentences for a specific year
  • query_SEC_sentences_by_sector(self, sector):
    • Finds companies by a sector and returns all of their risk-related SEC 10-K filing sentences
  • query_SEC_sentences_by_sector_year(self, sector, year):
    • Finds companies by a sector and returns all of their risk-related SEC 10-K filing sentences for a specific year
  • query_SEC_sentences_by_industry(self, industry):
    • Finds companies by an industry and returns all of their risk-related SEC 10-K filing sentences
  • query_SEC_sentences_by_industry_year(self, industry, year):
    • Finds companies by an industry and returns all of their risk-related SEC 10-K filing sentences for a specific year
  • query_by_risk_factor(self, risk_factor):
    • Returns all companies that have a specific risk factor

Query Assistance Functions

  • list_sectors(self):
  • list_industries(self):
  • list_companies(self):
  • list_risk_factors(self):

Export Functions

  • export_visualization(self) -> str:
    • Exports a .dot visualization of the graph
  • export_graph(self, output: str = "output.json", format_: str = "json-ld") -> str:
    • Exports the graph into specified formats

CLI.py

Description

  • Provides a command line interface for the Knowledge Graphs

      # Example Commands to Run
  quick_build
  add_company MSFT
  list_sectors
  list_industries
  list_companies
  list_risk_factors
  query_by_sector technology
  query_by_ticker IBM
  query_risk_factors AAPL
  query_by_risk_factor Energy
  query_SEC_sentences AIT
  query_SEC_sentences_by_year AIT 2010
  query_SEC_sentences_by_sector technology
  query_SEC_sentences_by_sector_year technology 2010
  export_graph

AttributeBuilder.py

Description

  • Provides functions that gets and loosely formats attribute information for the graph

Functions

  • def build_company(ticker: str) -> tuple[list[tuple[str, str]], str, tuple[str, str, str, str]]:
    • Gets information about a company from yfinance, loosely reformats it, and returns it
  • get_company_name(ticker: str) -> str:
    • Gets a company name from their ticker using yfinance
  • build_risk_factor(ticker: str) -> dict[str, tuple[list[str], list[tuple[str, str]]]]:
    • Extracts SEC 10-K filing sentences and related risk factors for a company, reformats it, and returns it

About

Languages

Language:Python 100.0%