@article{shirai-etal-2026-jfc,
    title = "{JFC}-Recipe: A Dataset for Nutrient Estimation from {J}apanese User-Generated Cooking Recipes",
    author = "Shirai, Keisuke  and
      Yamakata, Yoko  and
      Kameko, Hirotaka  and
      Sunto, Akiko  and
      Harashima, Jun  and
      Mori, Shinsuke",
    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.504/",
    pages = "6349--6360",
    abstract = "Estimating nutrients from recipes is essential for performing proper daily dietary control. The nutrients of the recipe could be roughly calculated by identifying the nutrients and weights of each ingredient in the recipe. However, no dataset with fully manual annotations of nutritional values and weights has been released so far, especially for Japanese recipes. In this work, we propose a novel dataset called the Japanese Food Composition Recipe Dataset (JFC-Recipe). The JFC-Recipe dataset consists of two types of annotations: (i) food item annotation that links ingredients in recipes to a database providing nutrients for foods and (ii) amount and unit annotation that are converted into weights in grams using a weight table. We describe a data collection procedure and annotation process, show statistics, and provide inter-annotator agreements to validate the quality of our annotations. In experiments, we tackle two tasks of food item estimation and quantity estimation. Experimental results show that pre-trained language models learn to estimate food items and quantities accurately."
}