SemEval-2023 Task 1 VWSD challenge training set
收藏arXiv2025-09-30 收录
下载链接：
https://semeval.github.io/SemEval2023/tasks.html
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含12,896个示例和13,000张候选图片，每个示例有10个候选图片，其中包括1张答案图片和9张干扰图片。每个示例的平均上下文长度为2.5个单词，且包含14.33%的词汇超出了数据集词汇表（在12,896个示例中有1,845个单词）。该数据集用于视觉词义消歧任务。
This dataset contains 12,896 examples and 13,000 candidate images. Each example includes 10 candidate images, comprising 1 target answer image and 9 distractor images. The average context length per example is 2.5 words, and 14.33% of the words across all 12,896 examples are out-of-vocabulary with respect to the dataset's vocabulary, with a total of 1,845 such out-of-vocabulary words. This dataset is developed for the visual word sense disambiguation task.
提供机构：
SemEval
