Scholarlink

<p>Credit Risk Assessment (CRA) is a critical process that evaluates a borrower’s creditworthiness and likelihood of default, particularly when extending credit to unknown applicants. Artificial Intelligence (AI) has introduced a wide array of tools that assist lenders in identifying high-risk borrowers with greater accuracy. This paper aims to examine the prominent AI techniques employed in both centralized (banking) and decentralized (peer-to-peer, P2P) lending models. Through a systematic analysis of 82 research articles published between 2014 and 2025, the review identifies key AI methodologies highlighting their application across various CRA phases such as data preprocessing, feature selection, credit scoring, and text mining, while also offering future insights to emerging trends and potential research directions. The paper also explores the role of behavioural data in banking and soft information in P2P lending as forms of alternative data that help address issues of information asymmetry and thin-file customers. It examines the ethical challenges associated with the use of alternative data, including legal concerns, data privacy, consent, security, and bias. The study proposes Responsible CRA framework to address these issues. Furthermore, customized AI-assisted data refinement processes are proposed to address data quality issues specific to both banking and P2P lending contexts.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A comparative review on AI empowered CRA process of banking and P2P lending models

  • 期刊论文

  • 发表日期：

    2026年3月27日

查看全文
  • Nisha Arora,
  • Pankaj Deep Kaur

摘要

Credit Risk Assessment (CRA) is a critical process that evaluates a borrower’s creditworthiness and likelihood of default, particularly when extending credit to unknown applicants. Artificial Intelligence (AI) has introduced a wide array of tools that assist lenders in identifying high-risk borrowers with greater accuracy. This paper aims to examine the prominent AI techniques employed in both centralized (banking) and decentralized (peer-to-peer, P2P) lending models. Through a systematic analysis of 82 research articles published between 2014 and 2025, the review identifies key AI methodologies highlighting their application across various CRA phases such as data preprocessing, feature selection, credit scoring, and text mining, while also offering future insights to emerging trends and potential research directions. The paper also explores the role of behavioural data in banking and soft information in P2P lending as forms of alternative data that help address issues of information asymmetry and thin-file customers. It examines the ethical challenges associated with the use of alternative data, including legal concerns, data privacy, consent, security, and bias. The study proposes Responsible CRA framework to address these issues. Furthermore, customized AI-assisted data refinement processes are proposed to address data quality issues specific to both banking and P2P lending contexts.