Legal Knowledge Graph (LegalKG)

A Legal Knowledge Graph (LegalKG) is a legal KB that is a domain-specific knowledge graph for legal concepts (legal entities and legal relations).

• Context:
  • It can be created by a LegalKG Creation Task (supported by a LegalKG creation system).
  • It can reference Legal Literature, including scientific literature and clinical data literature.
  • It can support Legal Tasks, such as: Contract Review, ....
  • It can support complex queries over connected Legal Data, and provide a structured representation of knowledge.
  • It can be associated to a Legal Ontology.
  • ...
• Example(s):
  • a Contract Knowledge Graph (such as the one given below by ChaptGPT-OpenAI, 2023),
  • a Legal Knowledge Extraction for Knowledge Graph based on a Question-Answering System (such as the one presented by Sovrano et al., 2020),
  • ...
• Counter-Example(s):
  • Government KG, such as OpenGov's KG.
  • Biomedical KG.
• See: Cinical Knowledge Extraction, Clinical Research KG.

---

---

References

2023

• (ChatGPT-OpenAI, 2023) ⇒ https://chat.openai.com Retrieved: 2023-09-10.
  • Q: Could you give a concrete example of a Legal Knowledge Graph (LegalKG)?
  • A: Certainly! A Legal Knowledge Graph (LegalKG) is a structured representation of legal concepts, entities, and their relationships in a graph format, which helps in organizing and navigating legal information. Here's a concrete example of a Legal Knowledge Graph:
    • Example: Contract Knowledge Graph

      In this example, we'll create a simplified Legal Knowledge Graph for a contract:

      • 1. Entities:
        • Parties: Nodes representing the parties involved in the contract, such as "Company A" and "Company B."
        • Clauses: Nodes representing specific contract clauses, like "Payment Terms", "Confidentiality", and "Termination".
        • Dates: Nodes representing important dates in the contract, such as "Effective Date" and "Expiration Date".
      • 2. Relationships:
        • Party A signs Contract with Party B: A directed edge connecting "Company A" to "Company B" to indicate their contractual relationship.
        • Contract contains Clause: Directed edges connecting the contract node to its clauses (e.g., "Contract" -> "Payment Terms").
        • Clause references Date: Directed edges connecting clauses to relevant dates (e.g., "Payment Terms" -> "Effective Date").
        • Clause references Parties: Directed edges connecting clauses to the parties involved (e.g., "Confidentiality" -> "Company A").
      • 3. Attributes:
        • Each node and edge in the graph can have attributes. For instance, the "Effective Date" node might have attributes like "Date: January 1, 2023," and the "Payment Terms" clause might have attributes like "Net 30 days".
      • 4. Queries and Analysis:
        • Users can query the graph to answer questions like:
          • "Which parties are involved in this contract?"
          • "What are the key clauses in the contract?"
          • "When does the contract expire?"
          • "Which clauses reference Company A?"
      • 5. Updates and Maintenance:
        • The graph can be continuously updated to reflect changes in the contract, such as amendments or revisions.

This Contract Knowledge Graph serves as a visual and structured representation of the contractual relationships, obligations, and key information within a legal document. It aids in contract analysis, due diligence, and understanding the legal implications of the agreement. Advanced legal knowledge graphs may incorporate more complex entities, legal precedents, and regulatory references to enhance their utility in legal research and compliance tasks.

2021

• (Github, 2021) ⇒ https://github.com/AnjaneyaTripathi/knowledge_graph
  • QUOTE: A knowledge graph has been constructed as a proof of concept on legal documents using the litigation releases from the SECs Website.

    The documents have been categorised into 4 main categories - Fraud, Insider Trading, Misappropriation of Funds and Unregistered Brokers. (There are more categories, but this is just a PoC so we have considered only these 4).

    After the classification, we have proceeded to extract relevant information from these litigations such as the violations, violators, action taken against the individuals as well as the fine imposed. These have been stored in a tabular format. The above data has also been prepared to be converted into a knowledge graph which is nested in nature.

2020

• (Sovrano et al., 2020) ⇒ Francesco Sovrano, Monica Palmirani, and Fabio Vitali. (2020). “Legal Knowledge Extraction for Knowledge Graph based Question-answering.” (PDF). In: Legal Knowledge and Information Systems, pp. 143-153 . IOS Press,
  • QUOTE: This paper presents a hybrid and innovative approach to model legal knowledge extracted from heterogeneous legal sources, using ontology design patterns as skeleton for mapping the information deducted using OKE and linguistic NLP analysis. In a legal domain, with multiple conflicting norms and a large number of multiple definitions for the same concept, our proposed approach gives interesting results, providing a KG where the legal expert can easily retrieve the relevant information via critical queries. The KG provides a useful instrument for information navigation, that could be integrated in traditional information systems and legal databases.(...)In the future we intend to integrate our approach in existing tools for legal document analysis, as the first part of a sophisticated explanatory tool for making sense of complex legal documents, facilitating the process of representing legal knowledge in machine-compatible ways (e.g. ontologies, taxonomies, thesauri, etc..).

---