MENU

FinCaKG: The Financial Causality Knowledge Graph for expertise discovery

FinCaKG: The Financial Causality Knowledge Graph for expertise discovery

Concept

Causality is an essential relation to revealing the logic of given contents. However, current causality knowledge graphs are not able to present the inner logic in a specific domain, i.e. finance. In this page, we present:

  • the methodology for FinCaKG construction
  • a video demonstration about a case study
  • the tutorial to access and interact with FinCaKG in Neo4j Browser remotely

To generate such a functional knowledge graph, we propose the methodology of causality detection module, entity linking module, and causality alignment module to automatically construct FinCaKG with the guidance of expert financial ontology - FIBO. Correspondingly, we draw the entire framework of constructing FinCaKG step by step:

Demonstration

Knowledge graph visualization. In the following demo, we demonstrate a case study of seeking for causes of a given keyword, i.e. "bad debt", using the visualization tool - Graphlytic. If the video is too small in your broswer, please click the upper-right button to pop out in a new tab.

FinCaKG in Neo4j Browser

You can access to our databases from Neo4j browser. Here are the instructions:

  1. Please click Neo4j interface link.
  2. Enter the following information to connect:
    3. Connect URL bolt://ec2-18-181-233-41.ap-northeast-1.compute.amazonaws.com:11002
    Database: neo4j://neo4j://ec2-18-181-233-41.ap-northeast-1.compute.amazonaws.com:11002
    Authentication type: Username/Password
    Username: neo4j
    Password: fincakg_2023
    Attention: After login, if not works well, please refresh your webpage.
  3. Now you can play with it. As shown in below, you can check the statistics and run the cypher clause to extract the dedicated sub-graphs.
    4. Attention: Please don't delete or alter the raw data. If you do it occationally, please wait for 24 hours to login again (we renew the database each 24 hours).

    MATCH path=(a)-[r:`fincakg:cause`]->(b) WHERE b.`rdfs:label` contains "Bad debt" return path

Download

Here are a list of different resources related to FinCaKG project.

FinCaKG dump

FinCaKG is composed by 127 nodes of FIBO_concept, 314 nodes of FIBO_instance and 1103 nodes of Investo_concept. Those nodes are connected by 32860 fincakf:cause relations and 134 taxonomic relations ( rdf:type and rdf:subClassOf).

We list the prefixes and namespaces used for FinCaKG ontology: 


          @prefix owl: <http://www.w3.org/2002/07/owl#> .
          @prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
          @prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
          @prefix dg: <https://w3id.org/dingo#> .
          @prefix fincakg: <https://www.ai.iee.e.titech.ac.jp/fincakg#> .
          @prefix fincakg-ind: <https://www.ai.iee.e.titech.ac.jp/fincakg/individuals#> .

Here is an example showing how to describe a node "Bad debt" in RDF/Turtle format: 


          fincakg-ind:2067 a fincakg:Investo_concept;
            rdfs:label "Bad debt";
            fincakg:occuredText "Bad Debt", "bad debt", "bad debts";
            fincakg:Class "Investo_concept" .
            rdfs:isDefinedBy "https://www.investopedia.com/terms/b/baddebt.asp"; 
            dg:wikidata_item_id "Q1365583";
            fincakg:occurAsEffect "0001381197-20-000006.767", "0001558370-20-001473.1483",...;
            fincakg:occurAsCause "0000929940-20-000069.1687", "0001564590-21-007047.126",...;

Also, we give an example showing how to associate the nodes "Bad debt" with 'fincakg:cause' relations in RDF/Turtle format: 


          fincakg-ind:908 a fincakg:Investo_concept;
            rdfs:label "Bankruptcy";
            fincakg:occuredText "bankruptcy", "Also included", "Bankruptcy", "bankruptcies";
            fincakg:Class "Investo_concept" .
            rdfs:isDefinedBy "https://www.investopedia.com/terms/b/bankruptcy.asp";
            dg:wikidata_item_id "Q152074";
            fincakg:occurAsEffect "0001709682-20-000017.69", "0000108385-20-000014.978",...;
            fincakg:occurAsCause "0001206264-21-000053.710", "0000215466-20-000032.3015",...;
            
            fincakg:cause fincakg-ind:2067;

Supplementary resources

We give random examples showing the concrete entity linking results, i.e. where key=101:

FIBO_concept in json format: 


    "101": {
      "label": [
          "service provider"
      ],
      "iri": "https://spec.edmcouncil.org/fibo/ontology/FND/ProductsAndServices/ProductsAndServices/ServiceProvider",
      "entity": [
          "Service provider",
          "en"
      ],
      "wikidata_id": "Q2169973"
      },

FIBO_instance in json format: 


    "101": {
      "label": "Afghānistān",
      "iri": "https://www.omg.org/spec/LCC/Countries/ISO3166-1-CountryCodes/Afghanistan",
      "entity": [
          "Afghanistan",
          "en"
      ],
      "wikidata_id": "Q889"
      },

Investo_concept in json format: 


    "101": {
      "raw_text": "Affiliate Marketing",
      "entity": [
          "Affiliate marketing",
          "en"
      ],
      "wikidata_id": "Q382453"
      },
Photograph is a derivative of work by FinCaKG.

Contact

Ziwei Xu: mail address
Ryutaro Ichise: mail address