@inproceedings{pan-etal-2025-negative,
    title = "Negative Matters: Multi-Granularity Hard-Negative Synthesis and Anchor-Token-Aware Pooling for Enhanced Text Embeddings",
    author = "Pan, Tengyu  and
      Duan, Zhichao  and
      Li, Zhenyu  and
      Dong, Bowen  and
      Liu, Ning  and
      Li, Xiuxing  and
      Wang, Jianyong",
    editor = "Che, Wanxiang  and
      Nabende, Joyce  and
      Shutova, Ekaterina  and
      Pilehvar, Mohammad Taher",
    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingestion-acl-25/2025.acl-long.1501/",
    pages = "31102--31118",
    ISBN = "979-8-89176-251-0",
    abstract = "Text embedding models are essential for various natural language processing tasks, enabling the effective encoding of semantic information into dense vector representations. These models are typically optimized using triplets of (query, positive, negative) data pairs for contrastive learning, where the negative samples play a critical role in enhancing the model{'}s ability to discern subtle semantic distinctions. In this work, we introduce a **M**ulti-**G**ranularity **H**ard-negative (MGH) synthesis framework that leverages large language models (LLMs) to generate diverse negative samples with varying levels of similarity with the query. This approach facilitates a coarse-to-fine curriculum learning strategy during supervised training, allowing the embedding model to progressively learn more nuanced semantic representations. Meanwhile, we propose an **A**nchor **T**oken **A**ware (ATA) pooling method that assigns higher weights to anchor tokens based on aggregation patterns observed in LLMs, improving text embedding accuracy without increasing model complexity. Comprehensive experiments on the MTEB benchmark demonstrate that our methods achieve state-of-the-art performance, surpassing existing synthesis strategies both with synthetic data and when combined with public retrieval datasets."
}