Identification of End-User Economical Relationship Graph Using Lightweight Blockchain-Based BERT Model

Area of Science:

• Natural Language Processing
• Information Extraction
• Network Analysis

Background:

• Current methods struggle with unstructured resumes in economic contexts.
• Extracting structured data and relationships from financial personnel resumes is challenging.

Purpose of the Study:

• To develop a framework for creating person relationship graphs in the economic sector.
• To convert unstructured user information into structured templates.
• To address limitations in current information extraction techniques for financial resumes.

Main Methods:

• Training a lightweight blockchain-based BERT model (B-BERT).
• Utilizing the B-BERT model for event instance vector extraction and categorization.
• Populating hierarchical user information templates with user characteristics.
• Generating person relationship graphs from structured templates.

Main Results:

• The proposed method efficiently retrieves text information from unstructured economic resumes.
• A character relationship graph in the economic sphere is successfully generated.
• The approach demonstrates effectiveness in building person relationship graphs.

Conclusions:

• The developed framework enables effective information extraction from unstructured financial resumes.
• The study successfully constructs person relationship graphs within the economic domain.
• This method offers a novel solution for analyzing relationships in financial personnel data.