ConFit v2: Improving Resume-Job Matching using Hypothetical Resume Embedding and Runner-Up Hard-Negative Mining

2025-02-17

AI Summary

In resume–job matching, extreme sparsity of interaction labels—arising from candidates applying to only a few positions—severely limits recommendation performance. To address this, we propose a contrastive learning framework that jointly integrates generative data augmentation and high-quality hard negative mining. Specifically, we leverage large language models (LLMs) to synthesize counterfactual positive-resume samples, alleviating the scarcity of ground-truth annotations; concurrently, we introduce a Runner-Up strategy to automatically identify semantically proximal hard negatives. Built upon a dual-tower encoder architecture, our method achieves state-of-the-art results on two real-world datasets, outperforming the prior best method ConFit by +13.8% in Recall@10 and +17.5% in nDCG@10—substantially surpassing both BM25 and OpenAI’s text-embedding-003.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

A reliable resume-job matching system helps a company recommend suitable candidates from a pool of resumes and helps a job seeker find relevant jobs from a list of job posts. However, since job seekers apply only to a few jobs, interaction labels in resume-job datasets are sparse. We introduce ConFit v2, an improvement over ConFit to tackle this sparsity problem. We propose two techniques to enhance the encoder's contrastive training process: augmenting job data with hypothetical reference resume generated by a large language model; and creating high-quality hard negatives from unlabeled resume/job pairs using a novel hard-negative mining strategy. We evaluate ConFit v2 on two real-world datasets and demonstrate that it outperforms ConFit and prior methods (including BM25 and OpenAI text-embedding-003), achieving an average absolute improvement of 13.8% in recall and 17.5% in nDCG across job-ranking and resume-ranking tasks.

Problem

• Enhances resume-job matching accuracy
• Addresses sparse interaction labels
• Improves contrastive training techniques

Innovation

• Hypothetical resume embedding
• Hard-negative mining strategy
• Contrastive training enhancement