Dataset Overview

Basic Information

• Language: English
• License: MIT
• Tags: knowledge-graph, rdf, owl, ontology
• Annotation Creator: Expert‑generated
• Dataset Name: FIBO
• Dataset Size: 100K<n<1M
• Task Category: graph‑ml

Dataset Features

• Features:
  • subject: string
  • predicate: string
  • object: string

Dataset Configurations

• Configuration Name: default
• Splits:
  • train:
    • Bytes: 56045523
    • Samples: 236579

Dataset Size

• Total Size: 56045523

Dataset Description

FIBO (Financial Industry Business Ontology) provides a structured framework for bridging the gap between theoretical financial concepts and real‑world data. The dataset consists of triples representing relationships between different financial concepts and named entities, such as market participants, companies and contract agents.

Use Cases

• Comprehensive Data Structure: FIBO covers a wide range of financial concepts from derivatives to securities.
• Decoding Complex Relationships: Complex inter‑dependencies in finance are clearly presented through FIBO’s structured approach.
• Linking to Real‑World Data: FIBO can associate financial concepts with real‑world financial data and controlled vocabularies.
• Enhanced Retrieval‑Augmented Generation: Large language models combined with RAG may revolutionize processing and interpretation of financial data.
• Document Classification: Using RAG to classify financial datasets by FIBO concepts helps analysts improve accuracy and depth of interpretation.

Construction and Validation

1. Construction: Imported from AboutFIBOProd‑IncludingReferenceData into Protégé 5.6.1.
2. Reasoning: Performed with the ELK reasoner plugin.
3. Consistency Check: Ensured ontology consistency using Protégé’s Debug Ontology plugin.
4. Export: After validation, exported inferred axioms, asserted axioms and annotations.
5. Encoding & Compression: Converted to N‑Triples with Apache Jena’s riot tool and compressed with gzip.

Usage

• Installation Requirements: python pip install datasets pip install rdflib

• Loading the Dataset: python from datasets import load_dataset dataset = load_dataset("wikipunk/fibo2023Q3", split="train")

Feature Description

• Subject: The subject of the triple, typically representing a specific financial instrument or entity.
• Predicate: The predicate, indicating the relationship between subject and object.
• Object: The object, the entity or value linked to the subject by the predicate.

Example

• Subject: <https://spec.edmcouncil.org/fibo/ontology/FBC/FunctionalEntities/MarketsIndividuals/ServiceProvider-L-JEUVK5RWVJEN8W0C9M24>
• Predicate: <http://www.w3.org/1999/02/22-rdf-syntax-ns#type>
• Object: <https://spec.edmcouncil.org/fibo/ontology/BE/FunctionalEntities/FunctionalEntities/FunctionalEntity>

Acknowledgements

We thank the FIBO contributors for their meticulous effort; their expertise and dedication are essential to shaping the innovative foundation of the financial industry.

Citation

bibtex @misc{fibo2023Q3, title={Financial Industry Business Ontology (FIBO)}, author={Object Management Group, Inc. and EDM Council, Inc. and Various Contributors}, year={2023}, note={Available as OWL 2 ontologies and UML models compliant with the Semantics for Information Modeling and Federation (SMIF) draft specification. Contributions are open on GitHub, consult the repository for a list of contributors.}, howpublished={url{https://spec.edmcouncil.org/fibo/}}, abstract={The Financial Industry Business Ontology (FIBO) is a collaborative effort to standardize the language used to define the terms, conditions, and characteristics of financial instruments; the legal and relationship structure of business entities; the content and time dimensions of market data; and the legal obligations and process aspects of corporate actions.}, license={MIT License, url{https://opensource.org/licenses/MIT}} }