@article{kaushik-etal-2026-aptfiner,
    title = "{APTF}i{NER}: Annotation Preserving Translation for Fine-grained Named Entity Recognition",
    author = "Kaushik, Prachuryya  and
      Gupta, Adittya  and
      Maurya, Ajanta  and
      Sharma, Gautam  and
      Saradhi, V. V.  and
      Anand, Ashish",
    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.608/",
    pages = "7668--7680",
    abstract = "We present APTFiNER, a novel fine-grained named entity recognition (FgNER) dataset covering six low-resource Indian languages spoken by over 400 million people across various nations. While creating FgNER resources through manual annotation is typically expensive and labor-intensive, distant supervision has emerged as a workable alternative. Yet, such FgNER datasets are often noisy, as each entity mentions are often assigned multiple entity types, which necessitates computationally demanding noise-aware models. Furthermore, resources for both coarse-grained and fine-grained NER tasks remain scarce for low-resource languages. To overcome this scarcity, we utilized the superior reasoning and translation capability of Gemini through the proposed annotation-preserving translation method and created a large-scale FgNER dataset comprising over 411 thousand sentences, 697 thousand entity mentions, and 5.8 million tokens in total. We translated the MultiCoNER2 English FgNER dataset to the target languages: {\ensuremath{<}}i{\ensuremath{>}}Assamese (as){\ensuremath{<}}/i{\ensuremath{>}}, {\ensuremath{<}}i{\ensuremath{>}}Marathi (mr){\ensuremath{<}}/i{\ensuremath{>}}, {\ensuremath{<}}i{\ensuremath{>}}Nepali (ne){\ensuremath{<}}/i{\ensuremath{>}}, {\ensuremath{<}}i{\ensuremath{>}}Tamil (ta){\ensuremath{<}}/i{\ensuremath{>}}, {\ensuremath{<}}i{\ensuremath{>}}Telugu (te){\ensuremath{<}}/i{\ensuremath{>}}, and a vulnerable language, {\ensuremath{<}}i{\ensuremath{>}}Bodo (brx){\ensuremath{<}}/i{\ensuremath{>}}. Through rigorous analyses and human evaluations, the effectiveness of our method and the high quality of the resulting dataset are ascertained with F1 score improvements of 8{\%} in both Tamil and Telugu, and 25{\%} in Marathi over the current state-of-the-art. The dataset, expert detector models, the agentic tool, and the interactive web application are available as open-source resources at: {\ensuremath{<}}url{\ensuremath{>}}https://hf.co/collections/prachuryyaIITG/aptfiner{\ensuremath{<}}/url{\ensuremath{>}}."
}