Scholarlink

<p>Online recruitment platforms play a critical role in enabling job seekers to access employment opportunities. However, many job seekers actively browse and apply for positions but receive little feedback, reflecting a fundamental challenge of unequal information access in digital labor markets, especially for information-disadvantaged users with sparse interaction histories and biased exposure. To address this challenge, this study proposes PA2Rec, a human-centric and generative AI enhanced framework that improves fair job information access through data-level augmentation. PA2Rec integrates a multi-attribute job filtering mechanism to construct high-quality candidate opportunity sets, leverages large language models to generate pseudo-interaction samples that expand users’ accessible opportunity space, and incorporates fairness-aware prompt tuning to mitigate systematic exposure bias. Experiments on the public Zhaopin (<a href="https://tianchi.aliyun.com/competition/entrance/231728/information">https://tianchi.aliyun.com/competition/entrance/231728/information</a>.) dataset demonstrate PA2Rec significantly improves recommendation accuracy and fairness, especially for information-disadvantaged job seekers. The results highlight the effectiveness of generative AI in augmenting information access and advancing human-centric recommendation systems in online recruitment.</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 Human-Centric Framework for Fair Job Recommendation Leveraging Generative AI and Data Augmentation

  • 期刊论文

  • 发表日期：

    2026年7月29日

查看全文
  • Jianshan Sun,
  • Jie Shao,
  • Kun Yuan,
  • Dongyang Li

摘要

Online recruitment platforms play a critical role in enabling job seekers to access employment opportunities. However, many job seekers actively browse and apply for positions but receive little feedback, reflecting a fundamental challenge of unequal information access in digital labor markets, especially for information-disadvantaged users with sparse interaction histories and biased exposure. To address this challenge, this study proposes PA2Rec, a human-centric and generative AI enhanced framework that improves fair job information access through data-level augmentation. PA2Rec integrates a multi-attribute job filtering mechanism to construct high-quality candidate opportunity sets, leverages large language models to generate pseudo-interaction samples that expand users’ accessible opportunity space, and incorporates fairness-aware prompt tuning to mitigate systematic exposure bias. Experiments on the public Zhaopin (https://tianchi.aliyun.com/competition/entrance/231728/information.) dataset demonstrate PA2Rec significantly improves recommendation accuracy and fairness, especially for information-disadvantaged job seekers. The results highlight the effectiveness of generative AI in augmenting information access and advancing human-centric recommendation systems in online recruitment.