# Similar Cases Recommendation using Legal Knowledge Graphs

**Authors:** Jaspreet Singh Dhani, Ruchika Bhatt, Balaji Ganesan, Parikshet Sirohi, Vasudha Bhatnagar  
**arXiv:** [2107.04771](https://arxiv.org/abs/2107.04771)  
**Date:** 2021-07-10  
**Summary page:** https://www.summarizepaper.com/en/paper/2107.04771v1/

## Key points

- A legal knowledge graph can revolutionize various applications in the field of law
- Utilizing knowledge graphs for downstream graph tasks like node similarity poses unique challenges
- The authors present an innovative solution for predicting similar nodes within a case graph using a legal knowledge graph
- Their approach aims to enhance the efficiency and effectiveness of legal research by providing recommendations based on similarities between different cases
- Previous research has focused on distant supervision in NLP tasks, but there is a scarcity of studies applying knowledge graphs to graph-based tasks like node similarity
- The authors propose a novel methodology that leverages their existing legal knowledge graph to predict similar nodes within a case graph
- Their solution's architecture can be utilized for recommending related cases or identifying relevant precedents
- This paper contributes to advancing the field of legal research and knowledge discovery by addressing challenges associated with selecting appropriate node types and defining relevant features for measuring similarity between nodes in a graph-based context.

## Summary

A legal knowledge graph constructed from court cases, judgments, laws, and other legal documents has the potential to revolutionize various applications in the field of law, such as question answering, document similarity analysis, and search. While the use of knowledge graphs for distant supervision in natural language processing (NLP) tasks is well-studied, utilizing knowledge graphs for downstream graph tasks like node similarity poses unique challenges in terms of selecting appropriate node types and their corresponding features. In this demo paper titled "Similar Cases Recommendation using Legal Knowledge Graphs," authors Jaspreet Singh Dhani, Ruchika Bhatt, Balaji Ganesan, Parikshet Sirohi, and Vasudha Bhatnagar present their innovative solution for predicting similar nodes within a case graph derived from a comprehensive legal knowledge graph. By leveraging the rich information contained within court cases, judgments, laws and other legal documents their approach aims to enhance the efficiency and effectiveness of legal research by providing recommendations based on similarities between different cases. The authors highlight that while previous research has extensively explored the use of knowledge graphs for distant supervision in NLP tasks there is a scarcity of studies focusing on applying knowledge graphs to graph-based tasks like node similarity. This gap arises due to the complexity involved in identifying relevant node types and determining suitable features for measuring similarity between nodes. To address this challenge the authors propose a novel methodology that leverages their existing legal knowledge graph to predict similar nodes within a case graph. They provide insights into their solution's architecture and discuss how it can be utilized to facilitate various applications such as recommending related cases or identifying relevant precedents. Overall this paper presents an innovative approach to harnessing the power of legal knowledge graphs for predicting similar nodes within a case graph. By addressing the challenges associated with selecting appropriate node types and defining relevant features for measuring similarity between nodes in a graph-based context this research contributes to advancing the field of legal research and knowledge discovery.