Results of the summarizing process for the arXiv paper: 2506.21607v1

Human smuggling networks are complex and constantly evolving, making them challenging to analyze. Legal case documents contain valuable insights but are often unstructured, dense with legal jargon, and filled with ambiguous references, posing obstacles for automated knowledge graph (KG) construction. Existing methods for building KGs lack coreference resolution and often result in noisy or fragmented graphs. To address these issues, a modular framework called CORE-KG has been proposed. This framework utilizes a two-step pipeline: type-aware coreference resolution through structured prompts and entity/relationship extraction guided by domain-specific instructions. To reduce noise and redundancy in constructing KGs that capture the structure of criminal networks from legal texts, targeted preprocessing is applied to extract only the "Opinion" section from each case file. This section contains the main factual narrative detailing individuals involved, routes used, items transported, and events sequence relevant to building meaningful KGs. In preliminary experiments using a sample of 20 cases retrieved from the Nexis Uni database, the system demonstrated its capacity to extract key entities and identify meaningful relationships. The implementation of LLaMA 3.3 70B model for coreference resolution and KG construction ensures reproducibility and minimizes stochastic variation during inference. The system operates within an open-source environment without relying on commercial APIs, enhancing accessibility and transparency. Baseline evaluation against GraphRAG framework highlights the effectiveness of CORE-KG in reducing node duplication by 33.28% and legal noise by 38.37%, resulting in cleaner and more coherent graph structures suitable for analyzing criminal networks like human smuggling. By focusing on extracting relevant information from the Opinion section of legal documents through coreference resolution and structured entity/relationship extraction processes, CORE-KG offers a comprehensive approach to constructing interpretable KGs from complex narratives found in criminal cases involving human smuggling activities.

• Human smuggling networks are complex and constantly evolving
• Legal case documents contain valuable insights but are often unstructured, dense with legal jargon, and filled with ambiguous references
• Existing methods for building knowledge graphs lack coreference resolution and often result in noisy or fragmented graphs
• A modular framework called CORE-KG has been proposed to address these issues
• CORE-KG utilizes a two-step pipeline: type-aware coreference resolution through structured prompts and entity/relationship extraction guided by domain-specific instructions
• Targeted preprocessing is applied to extract only the "Opinion" section from each case file to reduce noise and redundancy in constructing knowledge graphs
• In preliminary experiments using a sample of 20 cases, the system demonstrated its capacity to extract key entities and identify meaningful relationships
• The implementation of LLaMA 3.3 70B model for coreference resolution and KG construction ensures reproducibility and minimizes stochastic variation during inference
• The system operates within an open-source environment without relying on commercial APIs, enhancing accessibility and transparency
• Baseline evaluation against GraphRAG framework highlights the effectiveness of CORE-KG in reducing node duplication by 33.28% and legal noise by 38.37%
• By focusing on extracting relevant information from the Opinion section of legal documents through coreference resolution and structured entity/relationship extraction processes, CORE-KG offers a comprehensive approach to constructing interpretable KGs from complex narratives found in criminal cases involving human smuggling activities

Introduction

Human smuggling networks are a global phenomenon, with an estimated 5.9 million people being smuggled across international borders each year (UNODC, 2020). These networks are complex and constantly evolving, making them challenging to analyze. However, understanding the structure and operations of these networks is crucial for law enforcement agencies in their efforts to combat human smuggling.

One valuable source of information for analyzing human smuggling networks is legal case documents. These documents contain detailed narratives about individuals involved in the network, routes used for transportation, items transported, and events sequence relevant to building meaningful knowledge graphs (KGs). However, these documents are often unstructured and dense with legal jargon, making it difficult for automated systems to extract relevant information.

To address this issue, researchers have proposed a modular framework called CORE-KG (COREference Knowledge Graph) that utilizes a two-step pipeline: type-aware coreference resolution through structured prompts and entity/relationship extraction guided by domain-specific instructions. This article will provide a detailed overview of this research paper titled "CORE-KG: A Modular Framework for Constructing Interpretable Knowledge Graphs from Legal Case Documents on Human Smuggling Networks."

The Need for CORE-KG

Existing methods for constructing KGs lack coreference resolution capabilities and often result in noisy or fragmented graphs. Coreference resolution is the process of identifying all expressions in text that refer to the same real-world entity. In the context of human smuggling networks analysis, this means identifying all references to individuals involved in the network throughout the legal case document.

Without proper coreference resolution techniques, automated systems may mistakenly identify different references as separate entities or merge unrelated references into one entity. This can lead to noisy and inaccurate KGs that hinder effective analysis.

Furthermore, traditional methods do not take into account the specific domain knowledge required for extracting relevant information from legal case documents on human smuggling activities. This results in incomplete or irrelevant data being included in the KG, making it difficult to draw meaningful insights.

The CORE-KG Framework

To address these issues, the researchers proposed a modular framework called CORE-KG. This framework consists of two main components: type-aware coreference resolution and structured entity/relationship extraction.

Type-Aware Coreference Resolution

The first step in constructing an interpretable KG is to identify all references to entities involved in the human smuggling network. To achieve this, CORE-KG utilizes a type-aware coreference resolution approach that takes into account the specific types of entities relevant to human smuggling networks.

This is achieved through structured prompts that guide the system in identifying references to individuals involved in the network. These prompts are designed based on domain-specific knowledge and can be easily modified or extended as needed for different types of legal case documents.

Structured Entity/Relationship Extraction

Once all references have been identified through coreference resolution, the next step is to extract relevant information about these entities and their relationships from the legal case document. This process is guided by domain-specific instructions that specify which types of entities and relationships are relevant for building a meaningful KG for human smuggling networks analysis.

Targeted Preprocessing

To reduce noise and redundancy in constructing KGs from legal texts, targeted preprocessing is applied before extracting information from each case file. Specifically, only the "Opinion" section of each case file is extracted as it contains the main factual narrative detailing individuals involved, routes used, items transported, and events sequence relevant to building meaningful KGs.

Preliminary Experiments and Results

In preliminary experiments using a sample of 20 cases retrieved from the Nexis Uni database, CORE-KG demonstrated its capacity to extract key entities and identify meaningful relationships. The implementation of LLaMA 3.3 70B model for coreference resolution and KG construction ensures reproducibility and minimizes stochastic variation during inference.

Baseline evaluation against GraphRAG (Graph-based Reference-Aware Graph) framework highlights the effectiveness of CORE-KG in reducing node duplication by 33.28% and legal noise by 38.37%, resulting in cleaner and more coherent graph structures suitable for analyzing criminal networks like human smuggling.

Conclusion

In conclusion, the CORE-KG framework offers a comprehensive approach to constructing interpretable KGs from complex narratives found in criminal cases involving human smuggling activities. By focusing on extracting relevant information from the Opinion section of legal documents through coreference resolution and structured entity/relationship extraction processes, CORE-KG addresses key challenges faced by traditional methods and provides a valuable tool for law enforcement agencies in their efforts to combat human smuggling networks. Its open-source implementation also enhances accessibility and transparency, making it a promising solution for future research in this field.