@article{shalchian-etal-2026-persiananonymizer,
    title = "{P}ersian{A}nonymizer: Evaluating {LLM}-Labeled Training for Efficient {NER}-based Anonymization in {P}ersian",
    author = "Shalchian, Mohammad Hossein  and
      Amiri, Mostafa  and
      Sadeghzadeh, Amir Mahdi",
    editor = "Piperidis, Stelios  and
      Bel, N{\'u}ria  and
      van den Heuvel, Henk  and
      Ide, Nancy  and
      Krek, Simon  and
      Toral, Antonio",
    journal = "International Conference on Language Resources and Evaluation",
    volume = "main",
    month = may,
    year = "2026",
    address = "Palma de Mallorca, Spain",
    publisher = "ELRA Language Resource Association",
    url = "https://preview.aclanthology.org/ingest-lrec/2026.lrec-main.352/",
    pages = "4497--4506",
    abstract = "We target practical anonymization of Persian customer chats by training a compact NER model from LLM-labeled supervision and selecting the best labeler for deployment. We compare three instruction-tuned LLMs{---}DEEPSEEKV3-0324, GPT-OSS-120B, and QWEN3-235B-A22B-INSTRUCT-2507{---}to produce span annotations under a shared JSON protocol, yielding four corpora (OSS{\_}ZeroShot, Qwen{\_}ZeroShot, Qwen{\_}FewShot, DeepSeek{\_}FewShot). A MATINAROBERTA-based token-classifier is trained per corpus and evaluated with token-level Precision/Recall/F1 (overall and per-class). We also report Label Coverage Recall (LCR), the proportion of gold non-O tokens predicted as non-O, and quantify cross-labeler behavior via a token-level Venn on test annotations. Finally, we contrast test-set annotation latency of the LLMs on H200 nodes with the trained NER{'}s test-time labeling on a single RTX 3090. Results show that supervision from OSS{\_}ZeroShot yields the strongest macro-F1 and LCR, while the resulting NER labels an entire 40K-message test set in {\ensuremath{\sim}}2 minutes on one consumer GPU. This establishes a practical path to high-quality, low-cost anonymization for Persian industrial data."
}